@cursor/july 0.1.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/AGENTS.md +290 -0
- package/LICENSE.md +3 -0
- package/README.md +1233 -0
- package/dist/ab.d.ts +290 -0
- package/dist/ab.d.ts.map +1 -0
- package/dist/ab.js +386 -0
- package/dist/bin/agent-serve.d.ts +15 -0
- package/dist/bin/agent-serve.d.ts.map +1 -0
- package/dist/bin/agent-serve.js +754 -0
- package/dist/channels/github/api.d.ts +101 -0
- package/dist/channels/github/api.d.ts.map +1 -0
- package/dist/channels/github/api.js +229 -0
- package/dist/channels/github/auth.d.ts +67 -0
- package/dist/channels/github/auth.d.ts.map +1 -0
- package/dist/channels/github/auth.js +239 -0
- package/dist/channels/github/binding.d.ts +73 -0
- package/dist/channels/github/binding.d.ts.map +1 -0
- package/dist/channels/github/binding.js +114 -0
- package/dist/channels/github/defaults.d.ts +24 -0
- package/dist/channels/github/defaults.d.ts.map +1 -0
- package/dist/channels/github/defaults.js +100 -0
- package/dist/channels/github/events.d.ts +18 -0
- package/dist/channels/github/events.d.ts.map +1 -0
- package/dist/channels/github/events.js +59 -0
- package/dist/channels/github/forward.d.ts +137 -0
- package/dist/channels/github/forward.d.ts.map +1 -0
- package/dist/channels/github/forward.js +271 -0
- package/dist/channels/github/github-channel.d.ts +127 -0
- package/dist/channels/github/github-channel.d.ts.map +1 -0
- package/dist/channels/github/github-channel.js +1688 -0
- package/dist/channels/github/index.d.ts +52 -0
- package/dist/channels/github/index.d.ts.map +1 -0
- package/dist/channels/github/index.js +51 -0
- package/dist/channels/github/replay.d.ts +115 -0
- package/dist/channels/github/replay.d.ts.map +1 -0
- package/dist/channels/github/replay.js +282 -0
- package/dist/channels/github/state.d.ts +26 -0
- package/dist/channels/github/state.d.ts.map +1 -0
- package/dist/channels/github/state.js +50 -0
- package/dist/channels/github/types.d.ts +385 -0
- package/dist/channels/github/types.d.ts.map +1 -0
- package/dist/channels/github/types.js +1 -0
- package/dist/channels/slack/api.d.ts +117 -0
- package/dist/channels/slack/api.d.ts.map +1 -0
- package/dist/channels/slack/api.js +437 -0
- package/dist/channels/slack/approvals.d.ts +32 -0
- package/dist/channels/slack/approvals.d.ts.map +1 -0
- package/dist/channels/slack/approvals.js +157 -0
- package/dist/channels/slack/assistant.d.ts +23 -0
- package/dist/channels/slack/assistant.d.ts.map +1 -0
- package/dist/channels/slack/assistant.js +67 -0
- package/dist/channels/slack/auth.d.ts +40 -0
- package/dist/channels/slack/auth.d.ts.map +1 -0
- package/dist/channels/slack/auth.js +71 -0
- package/dist/channels/slack/channel-watch.d.ts +46 -0
- package/dist/channels/slack/channel-watch.d.ts.map +1 -0
- package/dist/channels/slack/channel-watch.js +180 -0
- package/dist/channels/slack/constants.d.ts +17 -0
- package/dist/channels/slack/constants.d.ts.map +1 -0
- package/dist/channels/slack/constants.js +27 -0
- package/dist/channels/slack/credentials.d.ts +35 -0
- package/dist/channels/slack/credentials.d.ts.map +1 -0
- package/dist/channels/slack/credentials.js +88 -0
- package/dist/channels/slack/cursor-account.d.ts +76 -0
- package/dist/channels/slack/cursor-account.d.ts.map +1 -0
- package/dist/channels/slack/cursor-account.js +88 -0
- package/dist/channels/slack/defaults.d.ts +32 -0
- package/dist/channels/slack/defaults.d.ts.map +1 -0
- package/dist/channels/slack/defaults.js +380 -0
- package/dist/channels/slack/dispatch.d.ts +43 -0
- package/dist/channels/slack/dispatch.d.ts.map +1 -0
- package/dist/channels/slack/dispatch.js +185 -0
- package/dist/channels/slack/doctor.d.ts +24 -0
- package/dist/channels/slack/doctor.d.ts.map +1 -0
- package/dist/channels/slack/doctor.js +115 -0
- package/dist/channels/slack/event-id-dedupe.d.ts +10 -0
- package/dist/channels/slack/event-id-dedupe.d.ts.map +1 -0
- package/dist/channels/slack/event-id-dedupe.js +23 -0
- package/dist/channels/slack/inbound.d.ts +41 -0
- package/dist/channels/slack/inbound.d.ts.map +1 -0
- package/dist/channels/slack/inbound.js +166 -0
- package/dist/channels/slack/index.d.ts +45 -0
- package/dist/channels/slack/index.d.ts.map +1 -0
- package/dist/channels/slack/index.js +44 -0
- package/dist/channels/slack/init.d.ts +32 -0
- package/dist/channels/slack/init.d.ts.map +1 -0
- package/dist/channels/slack/init.js +82 -0
- package/dist/channels/slack/interactive.d.ts +19 -0
- package/dist/channels/slack/interactive.d.ts.map +1 -0
- package/dist/channels/slack/interactive.js +142 -0
- package/dist/channels/slack/links.d.ts +10 -0
- package/dist/channels/slack/links.d.ts.map +1 -0
- package/dist/channels/slack/links.js +19 -0
- package/dist/channels/slack/live-delivery.d.ts +75 -0
- package/dist/channels/slack/live-delivery.d.ts.map +1 -0
- package/dist/channels/slack/live-delivery.js +143 -0
- package/dist/channels/slack/log.d.ts +25 -0
- package/dist/channels/slack/log.d.ts.map +1 -0
- package/dist/channels/slack/log.js +47 -0
- package/dist/channels/slack/manifest.d.ts +57 -0
- package/dist/channels/slack/manifest.d.ts.map +1 -0
- package/dist/channels/slack/manifest.js +110 -0
- package/dist/channels/slack/pr-ref.d.ts +18 -0
- package/dist/channels/slack/pr-ref.d.ts.map +1 -0
- package/dist/channels/slack/pr-ref.js +36 -0
- package/dist/channels/slack/setup.d.ts +14 -0
- package/dist/channels/slack/setup.d.ts.map +1 -0
- package/dist/channels/slack/setup.js +86 -0
- package/dist/channels/slack/slack-channel.d.ts +15 -0
- package/dist/channels/slack/slack-channel.d.ts.map +1 -0
- package/dist/channels/slack/slack-channel.js +281 -0
- package/dist/channels/slack/socket-mode.d.ts +26 -0
- package/dist/channels/slack/socket-mode.d.ts.map +1 -0
- package/dist/channels/slack/socket-mode.js +116 -0
- package/dist/channels/slack/stream.d.ts +26 -0
- package/dist/channels/slack/stream.d.ts.map +1 -0
- package/dist/channels/slack/stream.js +39 -0
- package/dist/channels/slack/tool-approval-events.d.ts +33 -0
- package/dist/channels/slack/tool-approval-events.d.ts.map +1 -0
- package/dist/channels/slack/tool-approval-events.js +129 -0
- package/dist/channels/slack/types.d.ts +326 -0
- package/dist/channels/slack/types.d.ts.map +1 -0
- package/dist/channels/slack/types.js +4 -0
- package/dist/channels.d.ts +116 -0
- package/dist/channels.d.ts.map +1 -0
- package/dist/channels.js +294 -0
- package/dist/connections.d.ts +75 -0
- package/dist/connections.d.ts.map +1 -0
- package/dist/connections.js +75 -0
- package/dist/docs/404.html +22 -0
- package/dist/docs/ab.html +87 -0
- package/dist/docs/assets/ab.md.COdXkces.js +63 -0
- package/dist/docs/assets/ab.md.COdXkces.lean.js +1 -0
- package/dist/docs/assets/app.DqfFEmJd.js +1 -0
- package/dist/docs/assets/building-with-agents.md.CJCtZCyi.js +13 -0
- package/dist/docs/assets/building-with-agents.md.CJCtZCyi.lean.js +1 -0
- package/dist/docs/assets/chunks/@localSearchIndexroot.CcVk1uKq.js +1 -0
- package/dist/docs/assets/chunks/VPLocalSearchBox.BaLEdS15.js +9 -0
- package/dist/docs/assets/chunks/framework.CAZyNGu9.js +19 -0
- package/dist/docs/assets/chunks/theme.CZRvu_0q.js +2 -0
- package/dist/docs/assets/concepts.md.Cfb9b-k1.js +4 -0
- package/dist/docs/assets/concepts.md.Cfb9b-k1.lean.js +1 -0
- package/dist/docs/assets/deployment.md.Dx1TYNk5.js +14 -0
- package/dist/docs/assets/deployment.md.Dx1TYNk5.lean.js +1 -0
- package/dist/docs/assets/evals.md.DPZ_MAnI.js +85 -0
- package/dist/docs/assets/evals.md.DPZ_MAnI.lean.js +1 -0
- package/dist/docs/assets/guides_agent-to-agent.md.CrtrsySy.js +10 -0
- package/dist/docs/assets/guides_agent-to-agent.md.CrtrsySy.lean.js +1 -0
- package/dist/docs/assets/guides_cloud-runtime.md.CYlNMTNp.js +9 -0
- package/dist/docs/assets/guides_cloud-runtime.md.CYlNMTNp.lean.js +1 -0
- package/dist/docs/assets/guides_github.md.DwbKhCeS.js +20 -0
- package/dist/docs/assets/guides_github.md.DwbKhCeS.lean.js +1 -0
- package/dist/docs/assets/guides_human-in-the-loop.md.Dvuctx7s.js +28 -0
- package/dist/docs/assets/guides_human-in-the-loop.md.Dvuctx7s.lean.js +1 -0
- package/dist/docs/assets/guides_slack.md.bv41fHfW.js +37 -0
- package/dist/docs/assets/guides_slack.md.bv41fHfW.lean.js +1 -0
- package/dist/docs/assets/guides_webhooks.md.hFTik3lf.js +50 -0
- package/dist/docs/assets/guides_webhooks.md.hFTik3lf.lean.js +1 -0
- package/dist/docs/assets/hillclimbing.md.DHNast08.js +4 -0
- package/dist/docs/assets/hillclimbing.md.DHNast08.lean.js +1 -0
- package/dist/docs/assets/index.md.BPKcj5AI.js +20 -0
- package/dist/docs/assets/index.md.BPKcj5AI.lean.js +1 -0
- package/dist/docs/assets/inter-italic-cyrillic-ext.r48I6akx.woff2 +0 -0
- package/dist/docs/assets/inter-italic-cyrillic.By2_1cv3.woff2 +0 -0
- package/dist/docs/assets/inter-italic-greek-ext.1u6EdAuj.woff2 +0 -0
- package/dist/docs/assets/inter-italic-greek.DJ8dCoTZ.woff2 +0 -0
- package/dist/docs/assets/inter-italic-latin-ext.CN1xVJS-.woff2 +0 -0
- package/dist/docs/assets/inter-italic-latin.C2AdPX0b.woff2 +0 -0
- package/dist/docs/assets/inter-italic-vietnamese.BSbpV94h.woff2 +0 -0
- package/dist/docs/assets/inter-roman-cyrillic-ext.BBPuwvHQ.woff2 +0 -0
- package/dist/docs/assets/inter-roman-cyrillic.C5lxZ8CY.woff2 +0 -0
- package/dist/docs/assets/inter-roman-greek-ext.CqjqNYQ-.woff2 +0 -0
- package/dist/docs/assets/inter-roman-greek.BBVDIX6e.woff2 +0 -0
- package/dist/docs/assets/inter-roman-latin-ext.4ZJIpNVo.woff2 +0 -0
- package/dist/docs/assets/inter-roman-latin.Di8DUHzh.woff2 +0 -0
- package/dist/docs/assets/inter-roman-vietnamese.BjW4sHH5.woff2 +0 -0
- package/dist/docs/assets/quickstart.md.tVPiGK_L.js +35 -0
- package/dist/docs/assets/quickstart.md.tVPiGK_L.lean.js +1 -0
- package/dist/docs/assets/reference_agent-config.md.Bpd7HQwf.js +34 -0
- package/dist/docs/assets/reference_agent-config.md.Bpd7HQwf.lean.js +1 -0
- package/dist/docs/assets/reference_channels.md.D7JTR03W.js +53 -0
- package/dist/docs/assets/reference_channels.md.D7JTR03W.lean.js +1 -0
- package/dist/docs/assets/reference_cli.md.DA730zCu.js +33 -0
- package/dist/docs/assets/reference_cli.md.DA730zCu.lean.js +1 -0
- package/dist/docs/assets/reference_connections.md.C3vNH_DE.js +26 -0
- package/dist/docs/assets/reference_connections.md.C3vNH_DE.lean.js +1 -0
- package/dist/docs/assets/reference_hooks.md.BCEc3MyM.js +12 -0
- package/dist/docs/assets/reference_hooks.md.BCEc3MyM.lean.js +1 -0
- package/dist/docs/assets/reference_http-api.md.DBAahtdz.js +11 -0
- package/dist/docs/assets/reference_http-api.md.DBAahtdz.lean.js +1 -0
- package/dist/docs/assets/reference_instructions.md.BC05LEQ8.js +12 -0
- package/dist/docs/assets/reference_instructions.md.BC05LEQ8.lean.js +1 -0
- package/dist/docs/assets/reference_playground.md.CPZhfYaO.js +1 -0
- package/dist/docs/assets/reference_playground.md.CPZhfYaO.lean.js +1 -0
- package/dist/docs/assets/reference_project-layout.md.Bd_CKtNS.js +31 -0
- package/dist/docs/assets/reference_project-layout.md.Bd_CKtNS.lean.js +1 -0
- package/dist/docs/assets/reference_schedules.md.D7qijxLk.js +82 -0
- package/dist/docs/assets/reference_schedules.md.D7qijxLk.lean.js +1 -0
- package/dist/docs/assets/reference_sessions.md.DdjWabnN.js +8 -0
- package/dist/docs/assets/reference_sessions.md.DdjWabnN.lean.js +1 -0
- package/dist/docs/assets/reference_skills.md.VQnlBT3Q.js +15 -0
- package/dist/docs/assets/reference_skills.md.VQnlBT3Q.lean.js +1 -0
- package/dist/docs/assets/reference_subagents.md.CIRAVcPK.js +10 -0
- package/dist/docs/assets/reference_subagents.md.CIRAVcPK.lean.js +1 -0
- package/dist/docs/assets/reference_tools.md.DF5kwlt0.js +46 -0
- package/dist/docs/assets/reference_tools.md.DF5kwlt0.lean.js +1 -0
- package/dist/docs/assets/scaffolding-agents.md.CyYfWGdc.js +1 -0
- package/dist/docs/assets/scaffolding-agents.md.CyYfWGdc.lean.js +1 -0
- package/dist/docs/assets/style.Bj-IXa5W.css +1 -0
- package/dist/docs/assets/troubleshooting.md.DkovjS7h.js +1 -0
- package/dist/docs/assets/troubleshooting.md.DkovjS7h.lean.js +1 -0
- package/dist/docs/building-with-agents.html +37 -0
- package/dist/docs/concepts.html +28 -0
- package/dist/docs/deployment.html +38 -0
- package/dist/docs/evals.html +109 -0
- package/dist/docs/guides/agent-to-agent.html +34 -0
- package/dist/docs/guides/cloud-runtime.html +33 -0
- package/dist/docs/guides/github.html +44 -0
- package/dist/docs/guides/human-in-the-loop.html +52 -0
- package/dist/docs/guides/slack.html +61 -0
- package/dist/docs/guides/webhooks.html +74 -0
- package/dist/docs/hashmap.json +1 -0
- package/dist/docs/hillclimbing.html +28 -0
- package/dist/docs/index.html +44 -0
- package/dist/docs/quickstart.html +59 -0
- package/dist/docs/reference/agent-config.html +58 -0
- package/dist/docs/reference/channels.html +77 -0
- package/dist/docs/reference/cli.html +57 -0
- package/dist/docs/reference/connections.html +50 -0
- package/dist/docs/reference/hooks.html +36 -0
- package/dist/docs/reference/http-api.html +35 -0
- package/dist/docs/reference/instructions.html +36 -0
- package/dist/docs/reference/playground.html +25 -0
- package/dist/docs/reference/project-layout.html +55 -0
- package/dist/docs/reference/schedules.html +106 -0
- package/dist/docs/reference/sessions.html +32 -0
- package/dist/docs/reference/skills.html +39 -0
- package/dist/docs/reference/subagents.html +34 -0
- package/dist/docs/reference/tools.html +70 -0
- package/dist/docs/scaffolding-agents.html +25 -0
- package/dist/docs/troubleshooting.html +25 -0
- package/dist/docs/vp-icons.css +0 -0
- package/dist/evals.d.ts +312 -0
- package/dist/evals.d.ts.map +1 -0
- package/dist/evals.js +254 -0
- package/dist/hooks.d.ts +23 -0
- package/dist/hooks.d.ts.map +1 -0
- package/dist/hooks.js +23 -0
- package/dist/index.d.ts +71 -0
- package/dist/index.d.ts.map +1 -0
- package/dist/index.js +101 -0
- package/dist/internal/ab-collector.d.ts +42 -0
- package/dist/internal/ab-collector.d.ts.map +1 -0
- package/dist/internal/ab-collector.js +153 -0
- package/dist/internal/ab-fold.d.ts +36 -0
- package/dist/internal/ab-fold.d.ts.map +1 -0
- package/dist/internal/ab-fold.js +171 -0
- package/dist/internal/ab-snapshot.d.ts +70 -0
- package/dist/internal/ab-snapshot.d.ts.map +1 -0
- package/dist/internal/ab-snapshot.js +208 -0
- package/dist/internal/approval-gate.d.ts +34 -0
- package/dist/internal/approval-gate.d.ts.map +1 -0
- package/dist/internal/approval-gate.js +85 -0
- package/dist/internal/auth.d.ts +8 -0
- package/dist/internal/auth.d.ts.map +1 -0
- package/dist/internal/auth.js +27 -0
- package/dist/internal/authored-alias-hooks.d.ts +31 -0
- package/dist/internal/authored-alias-hooks.d.ts.map +1 -0
- package/dist/internal/authored-alias-hooks.js +43 -0
- package/dist/internal/authored-loaders.d.ts +25 -0
- package/dist/internal/authored-loaders.d.ts.map +1 -0
- package/dist/internal/authored-loaders.js +129 -0
- package/dist/internal/bounded-int.d.ts +11 -0
- package/dist/internal/bounded-int.d.ts.map +1 -0
- package/dist/internal/bounded-int.js +13 -0
- package/dist/internal/brand.d.ts +6 -0
- package/dist/internal/brand.d.ts.map +1 -0
- package/dist/internal/brand.js +17 -0
- package/dist/internal/chat-client.d.ts +19 -0
- package/dist/internal/chat-client.d.ts.map +1 -0
- package/dist/internal/chat-client.js +327 -0
- package/dist/internal/cli-ax.d.ts +58 -0
- package/dist/internal/cli-ax.d.ts.map +1 -0
- package/dist/internal/cli-ax.js +345 -0
- package/dist/internal/cli-cursor.d.ts +20 -0
- package/dist/internal/cli-cursor.d.ts.map +1 -0
- package/dist/internal/cli-cursor.js +227 -0
- package/dist/internal/cli-github.d.ts +47 -0
- package/dist/internal/cli-github.d.ts.map +1 -0
- package/dist/internal/cli-github.js +644 -0
- package/dist/internal/cli-slack.d.ts +18 -0
- package/dist/internal/cli-slack.d.ts.map +1 -0
- package/dist/internal/cli-slack.js +144 -0
- package/dist/internal/cloud-merge.d.ts +23 -0
- package/dist/internal/cloud-merge.d.ts.map +1 -0
- package/dist/internal/cloud-merge.js +28 -0
- package/dist/internal/cron.d.ts +17 -0
- package/dist/internal/cron.d.ts.map +1 -0
- package/dist/internal/cron.js +60 -0
- package/dist/internal/cursor/account-mcp.d.ts +95 -0
- package/dist/internal/cursor/account-mcp.d.ts.map +1 -0
- package/dist/internal/cursor/account-mcp.js +342 -0
- package/dist/internal/cursor/backend-client.d.ts +118 -0
- package/dist/internal/cursor/backend-client.d.ts.map +1 -0
- package/dist/internal/cursor/backend-client.js +477 -0
- package/dist/internal/cursor/credentials.d.ts +112 -0
- package/dist/internal/cursor/credentials.d.ts.map +1 -0
- package/dist/internal/cursor/credentials.js +329 -0
- package/dist/internal/cursor-event-relay.d.ts +111 -0
- package/dist/internal/cursor-event-relay.d.ts.map +1 -0
- package/dist/internal/cursor-event-relay.js +181 -0
- package/dist/internal/cursor-relay-core.d.ts +113 -0
- package/dist/internal/cursor-relay-core.d.ts.map +1 -0
- package/dist/internal/cursor-relay-core.js +364 -0
- package/dist/internal/cursor-slack-relay.d.ts +94 -0
- package/dist/internal/cursor-slack-relay.d.ts.map +1 -0
- package/dist/internal/cursor-slack-relay.js +171 -0
- package/dist/internal/describe-error.d.ts +3 -0
- package/dist/internal/describe-error.d.ts.map +1 -0
- package/dist/internal/describe-error.js +4 -0
- package/dist/internal/dev-mode.d.ts +13 -0
- package/dist/internal/dev-mode.d.ts.map +1 -0
- package/dist/internal/dev-mode.js +17 -0
- package/dist/internal/discovery.d.ts +55 -0
- package/dist/internal/discovery.d.ts.map +1 -0
- package/dist/internal/discovery.js +1427 -0
- package/dist/internal/distribution.d.ts +47 -0
- package/dist/internal/distribution.d.ts.map +1 -0
- package/dist/internal/distribution.js +82 -0
- package/dist/internal/docs-site.d.ts +28 -0
- package/dist/internal/docs-site.d.ts.map +1 -0
- package/dist/internal/docs-site.js +169 -0
- package/dist/internal/duration.d.ts +8 -0
- package/dist/internal/duration.d.ts.map +1 -0
- package/dist/internal/duration.js +35 -0
- package/dist/internal/eval-run-store.d.ts +81 -0
- package/dist/internal/eval-run-store.d.ts.map +1 -0
- package/dist/internal/eval-run-store.js +396 -0
- package/dist/internal/eval-runner.d.ts +76 -0
- package/dist/internal/eval-runner.d.ts.map +1 -0
- package/dist/internal/eval-runner.js +336 -0
- package/dist/internal/event-log.d.ts +51 -0
- package/dist/internal/event-log.d.ts.map +1 -0
- package/dist/internal/event-log.js +167 -0
- package/dist/internal/event-mapper.d.ts +46 -0
- package/dist/internal/event-mapper.d.ts.map +1 -0
- package/dist/internal/event-mapper.js +226 -0
- package/dist/internal/frontmatter.d.ts +12 -0
- package/dist/internal/frontmatter.d.ts.map +1 -0
- package/dist/internal/frontmatter.js +61 -0
- package/dist/internal/github-fanout.d.ts +32 -0
- package/dist/internal/github-fanout.d.ts.map +1 -0
- package/dist/internal/github-fanout.js +122 -0
- package/dist/internal/handleAgentServeTrigger.d.ts +59 -0
- package/dist/internal/handleAgentServeTrigger.d.ts.map +1 -0
- package/dist/internal/handleAgentServeTrigger.js +115 -0
- package/dist/internal/host-platforms.d.ts +40 -0
- package/dist/internal/host-platforms.d.ts.map +1 -0
- package/dist/internal/host-platforms.js +113 -0
- package/dist/internal/http-channel.d.ts +54 -0
- package/dist/internal/http-channel.d.ts.map +1 -0
- package/dist/internal/http-channel.js +382 -0
- package/dist/internal/ids.d.ts +19 -0
- package/dist/internal/ids.d.ts.map +1 -0
- package/dist/internal/ids.js +34 -0
- package/dist/internal/init-project.d.ts +22 -0
- package/dist/internal/init-project.d.ts.map +1 -0
- package/dist/internal/init-project.js +125 -0
- package/dist/internal/json-dir-store.d.ts +32 -0
- package/dist/internal/json-dir-store.d.ts.map +1 -0
- package/dist/internal/json-dir-store.js +100 -0
- package/dist/internal/json-schema.d.ts +18 -0
- package/dist/internal/json-schema.d.ts.map +1 -0
- package/dist/internal/json-schema.js +53 -0
- package/dist/internal/log-ring.d.ts +23 -0
- package/dist/internal/log-ring.d.ts.map +1 -0
- package/dist/internal/log-ring.js +42 -0
- package/dist/internal/map-pool.d.ts +6 -0
- package/dist/internal/map-pool.d.ts.map +1 -0
- package/dist/internal/map-pool.js +41 -0
- package/dist/internal/mcp-endpoint.d.ts +59 -0
- package/dist/internal/mcp-endpoint.d.ts.map +1 -0
- package/dist/internal/mcp-endpoint.js +355 -0
- package/dist/internal/mcp-host.d.ts +28 -0
- package/dist/internal/mcp-host.d.ts.map +1 -0
- package/dist/internal/mcp-host.js +162 -0
- package/dist/internal/net.d.ts +5 -0
- package/dist/internal/net.d.ts.map +1 -0
- package/dist/internal/net.js +30 -0
- package/dist/internal/peer-connections.d.ts +44 -0
- package/dist/internal/peer-connections.d.ts.map +1 -0
- package/dist/internal/peer-connections.js +67 -0
- package/dist/internal/playground/static.d.ts +35 -0
- package/dist/internal/playground/static.d.ts.map +1 -0
- package/dist/internal/playground/static.js +162 -0
- package/dist/internal/playground/toolchain.d.ts +57 -0
- package/dist/internal/playground/toolchain.d.ts.map +1 -0
- package/dist/internal/playground/toolchain.js +226 -0
- package/dist/internal/reminder-runner.d.ts +62 -0
- package/dist/internal/reminder-runner.d.ts.map +1 -0
- package/dist/internal/reminder-runner.js +582 -0
- package/dist/internal/reminder-schedule.d.ts +22 -0
- package/dist/internal/reminder-schedule.d.ts.map +1 -0
- package/dist/internal/reminder-schedule.js +82 -0
- package/dist/internal/reminder-store.d.ts +58 -0
- package/dist/internal/reminder-store.d.ts.map +1 -0
- package/dist/internal/reminder-store.js +106 -0
- package/dist/internal/resolved-connections.d.ts +41 -0
- package/dist/internal/resolved-connections.d.ts.map +1 -0
- package/dist/internal/resolved-connections.js +73 -0
- package/dist/internal/review-comments.d.ts +50 -0
- package/dist/internal/review-comments.d.ts.map +1 -0
- package/dist/internal/review-comments.js +194 -0
- package/dist/internal/router.d.ts +33 -0
- package/dist/internal/router.d.ts.map +1 -0
- package/dist/internal/router.js +93 -0
- package/dist/internal/run-client.d.ts +49 -0
- package/dist/internal/run-client.d.ts.map +1 -0
- package/dist/internal/run-client.js +208 -0
- package/dist/internal/schedule-runner.d.ts +31 -0
- package/dist/internal/schedule-runner.d.ts.map +1 -0
- package/dist/internal/schedule-runner.js +126 -0
- package/dist/internal/sdk-runner.d.ts +41 -0
- package/dist/internal/sdk-runner.d.ts.map +1 -0
- package/dist/internal/sdk-runner.js +221 -0
- package/dist/internal/server.d.ts +25 -0
- package/dist/internal/server.d.ts.map +1 -0
- package/dist/internal/server.js +1374 -0
- package/dist/internal/session-engine.d.ts +287 -0
- package/dist/internal/session-engine.d.ts.map +1 -0
- package/dist/internal/session-engine.js +1377 -0
- package/dist/internal/session-store.d.ts +29 -0
- package/dist/internal/session-store.d.ts.map +1 -0
- package/dist/internal/session-store.js +140 -0
- package/dist/internal/session-urls.d.ts +10 -0
- package/dist/internal/session-urls.d.ts.map +1 -0
- package/dist/internal/session-urls.js +25 -0
- package/dist/internal/stream-progress.d.ts +14 -0
- package/dist/internal/stream-progress.d.ts.map +1 -0
- package/dist/internal/stream-progress.js +195 -0
- package/dist/internal/tool-result.d.ts +20 -0
- package/dist/internal/tool-result.d.ts.map +1 -0
- package/dist/internal/tool-result.js +45 -0
- package/dist/internal/trajectory.d.ts +56 -0
- package/dist/internal/trajectory.d.ts.map +1 -0
- package/dist/internal/trajectory.js +248 -0
- package/dist/internal/workspace.d.ts +42 -0
- package/dist/internal/workspace.d.ts.map +1 -0
- package/dist/internal/workspace.js +199 -0
- package/dist/playground/assets/cursor-icons-outline-oY2V_mvK.woff2 +0 -0
- package/dist/playground/assets/index-1K-hG-7p.css +1 -0
- package/dist/playground/assets/index-FlWjhg3x.js +79 -0
- package/dist/playground/index.html +17 -0
- package/dist/reminders.d.ts +23 -0
- package/dist/reminders.d.ts.map +1 -0
- package/dist/reminders.js +22 -0
- package/dist/schedules.d.ts +23 -0
- package/dist/schedules.d.ts.map +1 -0
- package/dist/schedules.js +26 -0
- package/dist/skills.d.ts +22 -0
- package/dist/skills.d.ts.map +1 -0
- package/dist/skills.js +22 -0
- package/dist/tools.d.ts +92 -0
- package/dist/tools.d.ts.map +1 -0
- package/dist/tools.js +68 -0
- package/dist/types.d.ts +1609 -0
- package/dist/types.d.ts.map +1 -0
- package/dist/types.js +80 -0
- package/docs/README.md +174 -0
- package/docs/ab.md +355 -0
- package/docs/building-with-agents.md +129 -0
- package/docs/concepts.md +177 -0
- package/docs/deployment.md +218 -0
- package/docs/evals.md +383 -0
- package/docs/guides/agent-to-agent.md +118 -0
- package/docs/guides/cloud-runtime.md +122 -0
- package/docs/guides/github.md +172 -0
- package/docs/guides/human-in-the-loop.md +131 -0
- package/docs/guides/slack.md +245 -0
- package/docs/guides/webhooks.md +201 -0
- package/docs/hillclimbing.md +93 -0
- package/docs/quickstart.md +183 -0
- package/docs/reference/agent-config.md +154 -0
- package/docs/reference/channels.md +206 -0
- package/docs/reference/cli.md +257 -0
- package/docs/reference/connections.md +137 -0
- package/docs/reference/hooks.md +74 -0
- package/docs/reference/http-api.md +219 -0
- package/docs/reference/instructions.md +77 -0
- package/docs/reference/playground.md +67 -0
- package/docs/reference/project-layout.md +119 -0
- package/docs/reference/schedules.md +220 -0
- package/docs/reference/sessions.md +165 -0
- package/docs/reference/skills.md +79 -0
- package/docs/reference/subagents.md +74 -0
- package/docs/reference/tools.md +203 -0
- package/docs/scaffolding-agents.md +113 -0
- package/docs/troubleshooting.md +95 -0
- package/package.json +185 -0
- package/skills/ab/SKILL.md +139 -0
- package/skills/create-agent/SKILL.md +219 -0
- package/skills/debug/SKILL.md +71 -0
- package/skills/evals/SKILL.md +217 -0
- package/skills/framework-map/SKILL.md +142 -0
- package/skills/github/SKILL.md +144 -0
- package/skills/hillclimb/SKILL.md +162 -0
- package/skills/setup-slack/SKILL.md +180 -0
- package/src/ab.ts +675 -0
- package/src/bin/agent-serve.ts +949 -0
- package/src/channels/github/api.ts +374 -0
- package/src/channels/github/auth.ts +275 -0
- package/src/channels/github/binding.ts +223 -0
- package/src/channels/github/defaults.ts +111 -0
- package/src/channels/github/events.ts +72 -0
- package/src/channels/github/forward.ts +381 -0
- package/src/channels/github/github-channel.ts +2270 -0
- package/src/channels/github/index.ts +52 -0
- package/src/channels/github/replay.ts +387 -0
- package/src/channels/github/state.ts +95 -0
- package/src/channels/github/types.ts +490 -0
- package/src/channels/slack/api.ts +580 -0
- package/src/channels/slack/approvals.ts +198 -0
- package/src/channels/slack/assistant.ts +76 -0
- package/src/channels/slack/auth.ts +112 -0
- package/src/channels/slack/channel-watch.ts +219 -0
- package/src/channels/slack/constants.ts +30 -0
- package/src/channels/slack/credentials.ts +90 -0
- package/src/channels/slack/cursor-account.ts +166 -0
- package/src/channels/slack/defaults.ts +450 -0
- package/src/channels/slack/dispatch.ts +293 -0
- package/src/channels/slack/doctor.ts +135 -0
- package/src/channels/slack/event-id-dedupe.ts +30 -0
- package/src/channels/slack/inbound.ts +219 -0
- package/src/channels/slack/index.ts +45 -0
- package/src/channels/slack/init.ts +133 -0
- package/src/channels/slack/interactive.ts +194 -0
- package/src/channels/slack/links.ts +23 -0
- package/src/channels/slack/live-delivery.ts +201 -0
- package/src/channels/slack/log.ts +83 -0
- package/src/channels/slack/manifest.ts +179 -0
- package/src/channels/slack/pr-ref.ts +50 -0
- package/src/channels/slack/setup.ts +97 -0
- package/src/channels/slack/slack-channel.ts +374 -0
- package/src/channels/slack/socket-mode.ts +154 -0
- package/src/channels/slack/stream.ts +63 -0
- package/src/channels/slack/tool-approval-events.ts +136 -0
- package/src/channels/slack/types.ts +349 -0
- package/src/channels.ts +385 -0
- package/src/connections.ts +89 -0
- package/src/evals.ts +501 -0
- package/src/hooks.ts +33 -0
- package/src/index.ts +184 -0
- package/src/internal/ab-collector.ts +210 -0
- package/src/internal/ab-fold.ts +228 -0
- package/src/internal/ab-snapshot.ts +335 -0
- package/src/internal/approval-gate.ts +115 -0
- package/src/internal/auth.ts +39 -0
- package/src/internal/authored-alias-hooks.ts +69 -0
- package/src/internal/authored-loaders.ts +138 -0
- package/src/internal/bounded-int.ts +27 -0
- package/src/internal/brand.ts +29 -0
- package/src/internal/chat-client.ts +407 -0
- package/src/internal/cli-ax.ts +474 -0
- package/src/internal/cli-cursor.ts +275 -0
- package/src/internal/cli-github.ts +849 -0
- package/src/internal/cli-slack.ts +175 -0
- package/src/internal/cloud-merge.ts +41 -0
- package/src/internal/cron.ts +78 -0
- package/src/internal/cursor/account-mcp.ts +472 -0
- package/src/internal/cursor/backend-client.ts +614 -0
- package/src/internal/cursor/credentials.ts +401 -0
- package/src/internal/cursor-event-relay.ts +281 -0
- package/src/internal/cursor-relay-core.ts +444 -0
- package/src/internal/cursor-slack-relay.ts +242 -0
- package/src/internal/describe-error.ts +4 -0
- package/src/internal/dev-mode.ts +20 -0
- package/src/internal/discovery.ts +1709 -0
- package/src/internal/distribution.ts +97 -0
- package/src/internal/docs-site.ts +191 -0
- package/src/internal/duration.ts +39 -0
- package/src/internal/eval-run-store.ts +519 -0
- package/src/internal/eval-runner.ts +522 -0
- package/src/internal/event-log.ts +175 -0
- package/src/internal/event-mapper.ts +272 -0
- package/src/internal/frontmatter.ts +77 -0
- package/src/internal/github-fanout.ts +159 -0
- package/src/internal/handleAgentServeTrigger.ts +156 -0
- package/src/internal/host-platforms.ts +144 -0
- package/src/internal/http-channel.ts +546 -0
- package/src/internal/ids.ts +44 -0
- package/src/internal/init-project.ts +165 -0
- package/src/internal/json-dir-store.ts +109 -0
- package/src/internal/json-schema.ts +69 -0
- package/src/internal/log-ring.ts +59 -0
- package/src/internal/map-pool.ts +32 -0
- package/src/internal/mcp-endpoint.ts +518 -0
- package/src/internal/mcp-host.ts +211 -0
- package/src/internal/net.ts +31 -0
- package/src/internal/peer-connections.ts +100 -0
- package/src/internal/playground/static.ts +189 -0
- package/src/internal/playground/toolchain.ts +277 -0
- package/src/internal/reminder-runner.ts +756 -0
- package/src/internal/reminder-schedule.ts +106 -0
- package/src/internal/reminder-store.ts +145 -0
- package/src/internal/resolved-connections.ts +145 -0
- package/src/internal/review-comments.ts +268 -0
- package/src/internal/router.ts +145 -0
- package/src/internal/run-client.ts +320 -0
- package/src/internal/schedule-runner.ts +161 -0
- package/src/internal/sdk-runner.ts +315 -0
- package/src/internal/server.ts +1737 -0
- package/src/internal/session-engine.ts +1810 -0
- package/src/internal/session-store.ts +147 -0
- package/src/internal/session-urls.ts +40 -0
- package/src/internal/stream-progress.ts +254 -0
- package/src/internal/tool-result.ts +54 -0
- package/src/internal/trajectory.ts +306 -0
- package/src/internal/workspace.ts +223 -0
- package/src/reminders.ts +35 -0
- package/src/schedules.ts +37 -0
- package/src/skills.ts +27 -0
- package/src/tools.ts +127 -0
- package/src/types.ts +1848 -0
|
@@ -0,0 +1,87 @@
|
|
|
1
|
+
<!DOCTYPE html>
|
|
2
|
+
<html lang="en-US" dir="ltr">
|
|
3
|
+
<head>
|
|
4
|
+
<meta charset="utf-8">
|
|
5
|
+
<meta name="viewport" content="width=device-width,initial-scale=1">
|
|
6
|
+
<title>Live A/B metrics | agentkit</title>
|
|
7
|
+
<meta name="description" content="Assign sticky variants with defineAB, collect live performance metrics, and inspect per-arm results.">
|
|
8
|
+
<meta name="generator" content="VitePress v1.6.4">
|
|
9
|
+
<link rel="preload stylesheet" href="/docs/assets/style.Bj-IXa5W.css" as="style">
|
|
10
|
+
<link rel="preload stylesheet" href="/docs/vp-icons.css" as="style">
|
|
11
|
+
|
|
12
|
+
<script type="module" src="/docs/assets/app.DqfFEmJd.js"></script>
|
|
13
|
+
<link rel="preload" href="/docs/assets/inter-roman-latin.Di8DUHzh.woff2" as="font" type="font/woff2" crossorigin="">
|
|
14
|
+
<link rel="modulepreload" href="/docs/assets/chunks/theme.CZRvu_0q.js">
|
|
15
|
+
<link rel="modulepreload" href="/docs/assets/chunks/framework.CAZyNGu9.js">
|
|
16
|
+
<link rel="modulepreload" href="/docs/assets/ab.md.COdXkces.lean.js">
|
|
17
|
+
<script id="check-dark-mode">(()=>{const e=localStorage.getItem("vitepress-theme-appearance")||"auto",a=window.matchMedia("(prefers-color-scheme: dark)").matches;(!e||e==="auto"?a:e==="dark")&&document.documentElement.classList.add("dark")})();</script>
|
|
18
|
+
<script id="check-mac-os">document.documentElement.classList.toggle("mac",/Mac|iPhone|iPod|iPad/i.test(navigator.platform));</script>
|
|
19
|
+
</head>
|
|
20
|
+
<body>
|
|
21
|
+
<div id="app"><div class="Layout" data-v-05085060><!--[--><!--]--><!--[--><span tabindex="-1" data-v-30dc752a></span><a href="#VPContent" class="VPSkipLink visually-hidden" data-v-30dc752a>Skip to content</a><!--]--><!----><header class="VPNav" data-v-05085060 data-v-cbe59881><div class="VPNavBar" data-v-cbe59881 data-v-dd8a1864><div class="wrapper" data-v-dd8a1864><div class="container" data-v-dd8a1864><div class="title" data-v-dd8a1864><div class="VPNavBarTitle has-sidebar" data-v-dd8a1864 data-v-f9346b2b><a class="title" href="/docs/" data-v-f9346b2b><!--[--><!--]--><!----><span data-v-f9346b2b>agentkit</span><!--[--><!--]--></a></div></div><div class="content" data-v-dd8a1864><div class="content-body" data-v-dd8a1864><!--[--><!--]--><div class="VPNavBarSearch search" data-v-dd8a1864><!--[--><!----><div id="local-search"><button type="button" class="DocSearch DocSearch-Button" aria-label="Search"><span class="DocSearch-Button-Container"><span class="vp-icon DocSearch-Search-Icon"></span><span class="DocSearch-Button-Placeholder">Search</span></span><span class="DocSearch-Button-Keys"><kbd class="DocSearch-Button-Key"></kbd><kbd class="DocSearch-Button-Key">K</kbd></span></button></div><!--]--></div><nav aria-labelledby="main-nav-aria-label" class="VPNavBarMenu menu" data-v-dd8a1864 data-v-7ed6e5ac><span id="main-nav-aria-label" class="visually-hidden" data-v-7ed6e5ac> Main Navigation </span><!--[--><!--[--><a class="VPLink link VPNavBarMenuLink" href="/docs/quickstart.html" tabindex="0" data-v-7ed6e5ac data-v-65720067><!--[--><span data-v-65720067>Quickstart</span><!--]--></a><!--]--><!--[--><a class="VPLink link VPNavBarMenuLink" href="/docs/guides/webhooks.html" tabindex="0" data-v-7ed6e5ac data-v-65720067><!--[--><span data-v-65720067>Guides</span><!--]--></a><!--]--><!--[--><a class="VPLink link VPNavBarMenuLink" href="/docs/reference/project-layout.html" tabindex="0" data-v-7ed6e5ac data-v-65720067><!--[--><span data-v-65720067>Reference</span><!--]--></a><!--]--><!--]--></nav><!----><div class="VPNavBarAppearance appearance" data-v-dd8a1864 data-v-2b2b395e><button class="VPSwitch VPSwitchAppearance" type="button" role="switch" title aria-checked="false" data-v-2b2b395e data-v-182deabf data-v-2562b538><span class="check" data-v-2562b538><span class="icon" data-v-2562b538><!--[--><span class="vpi-sun sun" data-v-182deabf></span><span class="vpi-moon moon" data-v-182deabf></span><!--]--></span></span></button></div><!----><div class="VPFlyout VPNavBarExtra extra" data-v-dd8a1864 data-v-f3b584b7 data-v-82133de6><button type="button" class="button" aria-haspopup="true" aria-expanded="false" aria-label="extra navigation" data-v-82133de6><span class="vpi-more-horizontal icon" data-v-82133de6></span></button><div class="menu" data-v-82133de6><div class="VPMenu" data-v-82133de6 data-v-759c9e19><!----><!--[--><!--[--><!----><div class="group" data-v-f3b584b7><div class="item appearance" data-v-f3b584b7><p class="label" data-v-f3b584b7>Appearance</p><div class="appearance-action" data-v-f3b584b7><button class="VPSwitch VPSwitchAppearance" type="button" role="switch" title aria-checked="false" data-v-f3b584b7 data-v-182deabf data-v-2562b538><span class="check" data-v-2562b538><span class="icon" data-v-2562b538><!--[--><span class="vpi-sun sun" data-v-182deabf></span><span class="vpi-moon moon" data-v-182deabf></span><!--]--></span></span></button></div></div></div><!----><!--]--><!--]--></div></div></div><!--[--><!--]--><button type="button" class="VPNavBarHamburger hamburger" aria-label="mobile navigation" aria-expanded="false" aria-controls="VPNavScreen" data-v-dd8a1864 data-v-ab07e844><span class="container" data-v-ab07e844><span class="top" data-v-ab07e844></span><span class="middle" data-v-ab07e844></span><span class="bottom" data-v-ab07e844></span></span></button></div></div></div></div><div class="divider" data-v-dd8a1864><div class="divider-line" data-v-dd8a1864></div></div></div><!----></header><div class="VPLocalNav has-sidebar empty" data-v-05085060 data-v-826b8761><div class="container" data-v-826b8761><button class="menu" aria-expanded="false" aria-controls="VPSidebarNav" data-v-826b8761><span class="vpi-align-left menu-icon" data-v-826b8761></span><span class="menu-text" data-v-826b8761>Menu</span></button><div class="VPLocalNavOutlineDropdown" style="--vp-vh:0px;" data-v-826b8761 data-v-01af9e73><button data-v-01af9e73>Return to top</button><!----></div></div></div><aside class="VPSidebar" data-v-05085060 data-v-5cd5b4d8><div class="curtain" data-v-5cd5b4d8></div><nav class="nav" id="VPSidebarNav" aria-labelledby="sidebar-aria-label" tabindex="-1" data-v-5cd5b4d8><span class="visually-hidden" id="sidebar-aria-label" data-v-5cd5b4d8> Sidebar Navigation </span><!--[--><!--]--><!--[--><div class="no-transition group" data-v-6b69f472><section class="VPSidebarItem level-0" data-v-6b69f472 data-v-b6893b1b><div class="item" role="button" tabindex="0" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><h2 class="text" data-v-b6893b1b>Getting started</h2><!----></div><div class="items" data-v-b6893b1b><!--[--><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Overview</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/quickstart.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Quickstart</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/scaffolding-agents.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Scaffold an agent with Cursor</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/concepts.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Concepts</p><!--]--></a><!----></div><!----></div><!--]--></div></section></div><div class="no-transition group" data-v-6b69f472><section class="VPSidebarItem level-0 has-active" data-v-6b69f472 data-v-b6893b1b><div class="item" role="button" tabindex="0" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><h2 class="text" data-v-b6893b1b>Improving agents</h2><!----></div><div class="items" data-v-b6893b1b><!--[--><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/building-with-agents.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Building agents with agents</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/evals.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Evals</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/ab.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Live A/B metrics</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/hillclimbing.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Hillclimbing</p><!--]--></a><!----></div><!----></div><!--]--></div></section></div><div class="no-transition group" data-v-6b69f472><section class="VPSidebarItem level-0" data-v-6b69f472 data-v-b6893b1b><div class="item" role="button" tabindex="0" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><h2 class="text" data-v-b6893b1b>Guides</h2><!----></div><div class="items" data-v-b6893b1b><!--[--><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/guides/webhooks.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Webhooks & custom channels</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/guides/github.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>GitHub</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/guides/slack.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Slack</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/guides/human-in-the-loop.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Human-in-the-loop approvals</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/guides/agent-to-agent.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Agent-to-agent</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/guides/cloud-runtime.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Cloud runtime</p><!--]--></a><!----></div><!----></div><!--]--></div></section></div><div class="no-transition group" data-v-6b69f472><section class="VPSidebarItem level-0" data-v-6b69f472 data-v-b6893b1b><div class="item" role="button" tabindex="0" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><h2 class="text" data-v-b6893b1b>Operating</h2><!----></div><div class="items" data-v-b6893b1b><!--[--><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/deployment.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Deployment</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/troubleshooting.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Fix common problems</p><!--]--></a><!----></div><!----></div><!--]--></div></section></div><div class="no-transition group" data-v-6b69f472><section class="VPSidebarItem level-0 collapsible" data-v-6b69f472 data-v-b6893b1b><div class="item" role="button" tabindex="0" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><h2 class="text" data-v-b6893b1b>Reference</h2><div class="caret" role="button" aria-label="toggle section" tabindex="0" data-v-b6893b1b><span class="vpi-chevron-right caret-icon" data-v-b6893b1b></span></div></div><div class="items" data-v-b6893b1b><!--[--><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/project-layout.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Project layout</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/agent-config.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Agent config</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/instructions.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Instructions</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/tools.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Tools</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/skills.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Skills</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/connections.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>MCP Connections</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/subagents.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Subagents</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/channels.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Channels</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/schedules.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Schedules & reminders</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/hooks.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Hooks</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/sessions.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Sessions & streaming</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/playground.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>Playground</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/cli.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>CLI</p><!--]--></a><!----></div><!----></div><div class="VPSidebarItem level-1 is-link" data-v-b6893b1b><div class="item" data-v-b6893b1b><div class="indicator" data-v-b6893b1b></div><a class="VPLink link link" href="/docs/reference/http-api.html" data-v-b6893b1b><!--[--><p class="text" data-v-b6893b1b>HTTP API</p><!--]--></a><!----></div><!----></div><!--]--></div></section></div><!--]--><!--[--><!--]--></nav></aside><div class="VPContent has-sidebar" id="VPContent" data-v-05085060 data-v-8eedf2ac><div class="VPDoc has-sidebar has-aside" data-v-8eedf2ac data-v-53b9fda6><!--[--><!--]--><div class="container" data-v-53b9fda6><div class="aside" data-v-53b9fda6><div class="aside-curtain" data-v-53b9fda6></div><div class="aside-container" data-v-53b9fda6><div class="aside-content" data-v-53b9fda6><div class="VPDocAside" data-v-53b9fda6 data-v-e448ecab><!--[--><!--]--><!--[--><!--]--><nav aria-labelledby="doc-outline-aria-label" class="VPDocAsideOutline" data-v-e448ecab data-v-b0b42374><div class="content" data-v-b0b42374><div class="outline-marker" data-v-b0b42374></div><div aria-level="2" class="outline-title" id="doc-outline-aria-label" role="heading" data-v-b0b42374>On this page</div><ul class="VPDocOutlineItem root" data-v-b0b42374 data-v-ae03bd52><!--[--><!--]--></ul></div></nav><!--[--><!--]--><div class="spacer" data-v-e448ecab></div><!--[--><!--]--><!----><!--[--><!--]--><!--[--><!--]--></div></div></div></div><div class="content" data-v-53b9fda6><div class="content-container" data-v-53b9fda6><!--[--><!--]--><main class="main" data-v-53b9fda6><div style="position:relative;" class="vp-doc _docs_ab" data-v-53b9fda6><div><h1 id="live-a-b-metrics" tabindex="-1">Live A/B metrics <a class="header-anchor" href="#live-a-b-metrics" aria-label="Permalink to "Live A/B metrics""></a></h1><p>Use <code>defineAB</code> to compare variants on live agent sessions. New sessions receive a sticky assignment in each enrolled experiment. agentkit folds their durable event streams into tool, token, failure, and wall-time metrics. You can send cumulative samples to your metrics backend and inspect aggregates in the playground.</p><p><code>defineAB</code> compares live variants through sticky assignment, instruction overlays, optional tool branches, and cumulative metrics. Metric callbacks observe the result without approving, rejecting, or failing a turn. Use <a href="./evals.html">evals</a> for pass/fail regression checks on fixed inputs.</p><div class="note custom-block github-alert"><p class="custom-block-title">NOTE</p><p>Import paths here use <code>@anysphere/agentkit/ab</code>. On projects still using <code>@anysphere/agent-serve</code>, swap the import. See <a href="./README.html#run-the-cli">Run the CLI</a> for the full rename table.</p></div><h2 id="choose-live-a-b-metrics-or-evals" tabindex="-1">Choose live A/B metrics or evals <a class="header-anchor" href="#choose-live-a-b-metrics-or-evals" aria-label="Permalink to "Choose live A/B metrics or evals""></a></h2><p>Both features read the session event stream, but they answer different questions.</p><table tabindex="0"><thead><tr><th></th><th>Live A/B metrics</th><th>Evals</th></tr></thead><tbody><tr><td>Question</td><td>How do variants compare on live sessions?</td><td>Does the agent still meet a fixed contract?</td></tr><tr><td>Location</td><td><code>agent/ab.ts</code> or <code>agent/ab/<name>.ts</code></td><td><code>evals/**/*.eval.ts</code></td></tr><tr><td>Input</td><td>Dev or production traffic</td><td>Frozen prompts and fixtures</td></tr><tr><td>Output</td><td>Cumulative metrics by session and arm</td><td>Pass/fail assertions</td></tr><tr><td>How it runs</td><td>Automatically on new live sessions</td><td><code>agentkit eval</code></td></tr></tbody></table><p>There is no <code>agentkit ab</code> command or assertion API.</p><h2 id="define-an-experiment" tabindex="-1">Define an experiment <a class="header-anchor" href="#define-an-experiment" aria-label="Permalink to "Define an experiment""></a></h2><p>Author one experiment in <code>agent/ab.ts</code>, add more under <code>agent/ab/<name>.ts</code>, or use both forms. Each file defines one experiment. The experiment name comes from <code>name</code> when set. Otherwise, agentkit uses <code>ab</code> for <code>agent/ab.ts</code> and the file stem for files under <code>agent/ab/</code>.</p><div class="language-ts vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">ts</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;">// agent/ab/concise-weather.ts</span></span>
|
|
22
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">import</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> {</span></span>
|
|
23
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> defineAB,</span></span>
|
|
24
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> splitBySessionHash,</span></span>
|
|
25
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">} </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">from</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "@anysphere/agentkit/ab"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">;</span></span>
|
|
26
|
+
<span class="line"></span>
|
|
27
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">export</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> default</span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> defineAB</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">({</span></span>
|
|
28
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> name: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"concise-weather"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
29
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> variants: {</span></span>
|
|
30
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> control: {</span></span>
|
|
31
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> label: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"Baseline"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
32
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
33
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> treatment: {</span></span>
|
|
34
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> label: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"Short replies"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
35
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> description: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"Adds a one-paragraph response limit."</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
36
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> instructions: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"Keep weather replies to one short paragraph."</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
37
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
38
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
39
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> split: </span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">splitBySessionHash</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">({</span></span>
|
|
40
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> weights: { control: </span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;">1</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">, treatment: </span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;">1</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
41
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> holdout: </span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;">0.1</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
42
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> }),</span></span>
|
|
43
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> derive: {</span></span>
|
|
44
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> weatherCalls</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">: (</span><span style="--shiki-light:#E36209;--shiki-dark:#FFAB70;">event</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">) </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">=></span></span>
|
|
45
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> event.type </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">===</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "action.result"</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> &&</span></span>
|
|
46
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> event.data.toolName </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">===</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "get_weather"</span></span>
|
|
47
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> ?</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> 1</span></span>
|
|
48
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> :</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> null</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
49
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
50
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> onSample</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span><span style="--shiki-light:#E36209;--shiki-dark:#FFAB70;">sample</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">) {</span></span>
|
|
51
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> console.</span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">log</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span></span>
|
|
52
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> sample.experiment,</span></span>
|
|
53
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> sample.variant,</span></span>
|
|
54
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> sample.metrics.toolCalls,</span></span>
|
|
55
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> sample.metrics.wallTimeMs</span></span>
|
|
56
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> );</span></span>
|
|
57
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
58
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">});</span></span></code></pre></div><p>Every definition needs:</p><ul><li>At least two variants. Variant keys cannot be empty or contain <code>/</code> or <code>\</code>.</li><li>A <code>split</code> function that returns a variant key or <code>null</code>.</li><li>An <code>onSample</code> callback for completed or failed turns.</li></ul><p><code>label</code> and <code>description</code> appear with the arm in result surfaces. <code>instructions</code> changes the prompt for sessions in that arm. <code>derive</code> adds custom counters.</p><p>Duplicate experiment names are validation errors. Check discovery before you serve:</p><div class="language-bash vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">bash</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> validate</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span></span>
|
|
59
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> info</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --json</span></span></code></pre></div><p>The <code>abs</code> field in <code>info</code> lists the discovered experiment names.</p><h2 id="assign-sticky-variants" tabindex="-1">Assign sticky variants <a class="header-anchor" href="#assign-sticky-variants" aria-label="Permalink to "Assign sticky variants""></a></h2><p>Enrollment happens once, when a live session is created and before its first turn:</p><ol><li>agentkit records <code>session.started</code>.</li><li>Each experiment runs its <code>split</code> function.</li><li>agentkit records one durable <code>ab.assigned</code> event per experiment.</li><li>The selected arms become available on <code>session.abs</code>.</li><li>Variant instruction overlays reach the first model turn.</li></ol><p>A split can return a variant key or <code>null</code>. A null assignment is a sticky skip for that experiment. It increments the experiment's <code>skipped</code> total, still appears in the snapshot's <code>sessions</code> list with <code>variant: null</code>, and does not collect arm metrics or call <code>onSample</code>.</p><p>Use the split helper that matches your rollout:</p><table tabindex="0"><thead><tr><th>Helper</th><th>Behavior</th></tr></thead><tbody><tr><td><code>splitBySessionHash({ weights?, holdout?, salt? })</code></td><td>Hashes the session id into a reproducible arm; the recommended default</td></tr><tr><td><code>splitByRandom({ weights?, holdout? })</code></td><td>Draws once when the session starts, then persists the result</td></tr><tr><td><code>splitAlways("control")</code></td><td>Pins every new session to one arm</td></tr><tr><td><code>splitNone()</code></td><td>Skips every new session without deleting the experiment</td></tr><tr><td><code>splitIf(predicate, inner)</code></td><td>Runs <code>inner</code> only when the predicate passes</td></tr><tr><td>Custom <code>split(ctx)</code></td><td>Returns a declared variant key or <code>null</code></td></tr></tbody></table><p>The split context includes the agent name, channel id, session info, experiment name, and declared variant keys. For example, enroll only Slack sessions:</p><div class="language-ts vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">ts</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">split</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">: </span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">splitIf</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span></span>
|
|
60
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> (</span><span style="--shiki-light:#E36209;--shiki-dark:#FFAB70;">ctx</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">) </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">=></span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> ctx.channel.id </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">===</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "slack"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
61
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> splitBySessionHash</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">()</span></span>
|
|
62
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">),</span></span></code></pre></div><p>Weights default to equal. Non-positive weights leave an arm out of the draw, and at least one arm must have a positive weight. <code>holdout</code> is the fraction of sessions assigned <code>null</code>, from <code>0</code> through <code>1</code>. Change <code>salt</code> to reshuffle future hash assignments without renaming the experiment.</p><p>If a custom split throws or returns an unknown variant, agentkit logs the error and records <code>variant: null</code>. The failed decision becomes a sticky skip instead of breaking the session.</p><p>Enrollment only applies to new sessions. Adding an experiment does not assign existing conversations. Follow-ups keep the session's original arms. Keep experiment names and variant keys stable while you collect and compare results.</p><h2 id="change-behavior-by-variant" tabindex="-1">Change behavior by variant <a class="header-anchor" href="#change-behavior-by-variant" aria-label="Permalink to "Change behavior by variant""></a></h2><p>Variant instructions are appended to the agent's base instructions. Local sessions receive the merged instructions in <code>AGENTS.md</code> before every turn. Cloud sessions receive them in the first-turn preamble only. For cloud follow-ups, branch through <code>session.abs</code> when the arm must remain visible to deterministic behavior.</p><p>Tools can branch on the assignment through <code>ctx.session.abs</code>. Hooks can read the same field for logging or export:</p><div class="language-ts vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">ts</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">const</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> treatment</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> =</span></span>
|
|
63
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> ctx.session.abs?.[</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"concise-weather"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">] </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">===</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "treatment"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">;</span></span>
|
|
64
|
+
<span class="line"></span>
|
|
65
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">if</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> (treatment) {</span></span>
|
|
66
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> return</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> conciseWeatherResult;</span></span>
|
|
67
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">}</span></span>
|
|
68
|
+
<span class="line"></span>
|
|
69
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">return</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> baselineWeatherResult;</span></span></code></pre></div><p>This makes the assignment available to deterministic code as well as the model prompt. Use both patterns together when one experiment must steer the prompt and host code at once.</p><p><code>defineAB</code> does not select a different model or runtime for each arm. Keep those settings in <code>agent/agent.ts</code>, or write explicit host logic when your experiment needs another behavior lever.</p><p>The split and selected arm can affect agent behavior. <code>derive</code> and <code>onSample</code> only observe the resulting event stream. Errors in either callback are logged and never fail the turn.</p><h2 id="collect-built-in-and-custom-metrics" tabindex="-1">Collect built-in and custom metrics <a class="header-anchor" href="#collect-built-in-and-custom-metrics" aria-label="Permalink to "Collect built-in and custom metrics""></a></h2><p>Metrics accumulate for each session and experiment. When one session joins several experiments, every enrolled experiment folds the same turn and tool events into its own counters.</p><table tabindex="0"><thead><tr><th>Metric</th><th>How agentkit calculates it</th></tr></thead><tbody><tr><td><code>turns</code></td><td>Adds one on <code>turn.completed</code> or <code>turn.failed</code></td></tr><tr><td><code>turnFailures</code></td><td>Adds one on <code>turn.failed</code></td></tr><tr><td><code>toolCalls</code></td><td>Adds one for each <code>action.result</code></td></tr><tr><td><code>toolErrors</code></td><td>Adds one when <code>action.result.data.isError</code> is true</td></tr><tr><td><code>inputTokens</code>, <code>outputTokens</code></td><td>Adds usage from completed turns</td></tr><tr><td><code>cacheReadTokens</code>, <code>cacheWriteTokens</code></td><td>Adds cache usage from completed turns</td></tr><tr><td><code>wallTimeMs</code></td><td>Sums the time from <code>turn.started</code> to its completed or failed event</td></tr><tr><td><code>custom</code></td><td>Sums finite numeric deltas returned by <code>derive</code></td></tr></tbody></table><p><code>onSample</code> fires after every <code>turn.completed</code> and <code>turn.failed</code> event for an enrolled arm. The sample contains:</p><table tabindex="0"><thead><tr><th>Field</th><th>Value</th></tr></thead><tbody><tr><td><code>experiment</code></td><td>Experiment name</td></tr><tr><td><code>variant</code>, <code>variantLabel?</code></td><td>Sticky arm and optional display label</td></tr><tr><td><code>sessionId</code>, <code>channelId</code></td><td>Source session</td></tr><tr><td><code>metrics</code></td><td>Cumulative metrics through this turn</td></tr><tr><td><code>reason</code></td><td><code>turn.completed</code> or <code>turn.failed</code></td></tr><tr><td><code>at</code></td><td>Terminal event timestamp</td></tr></tbody></table><p>The metrics are cumulative, not per-turn deltas. A second sample from the same session includes the first turn's counts.</p><p>Each <code>derive</code> extractor runs on every session event for its enrolled experiment, including streamed <code>message.appended</code> events. Keep it synchronous and cheap. Return a finite number to add a delta, or <code>null</code> to skip the event. Send samples to your metrics service from <code>onSample</code>; do not perform network or disk work in <code>derive</code>.</p><p>Skipped sessions never call <code>onSample</code>. Errors from <code>derive</code> or <code>onSample</code> are logged, then metric collection continues.</p><h2 id="inspect-assignments-and-results" tabindex="-1">Inspect assignments and results <a class="header-anchor" href="#inspect-assignments-and-results" aria-label="Permalink to "Inspect assignments and results""></a></h2><p>Open the playground's <strong>A/Bs</strong> tab to see aggregate arm totals and per-session assignments. The tab reads <code>GET /v1/abs</code>.</p><p>The response has two views of the same durable data:</p><table tabindex="0"><thead><tr><th>Field</th><th>Contents</th></tr></thead><tbody><tr><td><code>experiments</code></td><td>Declared variants, skipped-session count, arm session counts, and aggregate metrics</td></tr><tr><td><code>sessions</code></td><td>Visible sessions with their assignments and cumulative metrics</td></tr></tbody></table><p><code>GET /v1/abs</code> returns sessions visible to the current principal by default. In <code>--dev</code>, loopback requests include every session. Add <code>--allow-anonymous</code> to include every session from non-loopback callers too. This include-all behavior can still apply to <code>GET /v1/abs</code> in dev when bearer or custom auth keeps <code>GET /v1/sessions</code> owner-scoped.</p><p>Session <code>events.ndjson</code> is the source of truth for assignment + fold. <code>GET /v1/abs</code> recomputes aggregates from those logs. Use <code>agent/ab.config.ts</code> (and/or each experiment's <code>onSample</code>) when you want author-controlled sample/snapshot exports — same idea as eval <code>persistRuns</code>.</p><h2 id="configure-retention-and-persistence" tabindex="-1">Configure retention and persistence <a class="header-anchor" href="#configure-retention-and-persistence" aria-label="Permalink to "Configure retention and persistence""></a></h2><p>Optional project defaults in <code>agent/ab.config.ts</code>:</p><div class="language-ts vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">ts</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">import</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> {</span></span>
|
|
70
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> defineABConfig,</span></span>
|
|
71
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> persistABSamplesToDir,</span></span>
|
|
72
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> persistABSnapshotsToDir,</span></span>
|
|
73
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">} </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">from</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "@anysphere/agentkit/ab"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">;</span></span>
|
|
74
|
+
<span class="line"></span>
|
|
75
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">export</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> default</span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> defineABConfig</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">({</span></span>
|
|
76
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // Optional — defaults to 200. Only affects GET /v1/abs / A/Bs tab.</span></span>
|
|
77
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // maxPlaygroundSessions: 200,</span></span>
|
|
78
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // retainSnapshots: 20,</span><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // prune budget for persistABSnapshotsToDir</span></span>
|
|
79
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // Append onSample payloads (in addition to each experiment's onSample)</span></span>
|
|
80
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> persistSamples: </span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">persistABSamplesToDir</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">".agent-serve/ab-samples"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">),</span></span>
|
|
81
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // Save aggregate snapshots whenever GET /v1/abs runs</span></span>
|
|
82
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> persistSnapshots: </span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">persistABSnapshotsToDir</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">".agent-serve/ab-snapshots"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">),</span></span>
|
|
83
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">});</span></span></code></pre></div><table tabindex="0"><thead><tr><th>Option</th><th>Default</th><th>Meaning</th></tr></thead><tbody><tr><td><code>maxPlaygroundSessions</code></td><td><code>200</code></td><td>Max newest sessions folded into the playground / <code>GET /v1/abs</code> (not session logs or <code>persistSamples</code>)</td></tr><tr><td><code>persistSamples</code></td><td>unset</td><td>Durable sink for metric samples (<code>save(sample, ctx)</code>)</td></tr><tr><td><code>persistSnapshots</code></td><td>unset</td><td>Durable store for aggregate snapshots (<code>load</code> / <code>save</code> / optional <code>delete</code>)</td></tr><tr><td><code>retainSnapshots</code></td><td><code>20</code> when snapshots persist</td><td>Snapshot files kept after prune (<code>persistABSnapshotsToDir</code> honors this)</td></tr></tbody></table><p>Implement your own <code>{ save }</code> / <code>{ load, save, delete? }</code> for S3, a DB, or your metrics vendor. Without <code>persistSamples</code> / <code>persistSnapshots</code>, samples only go where each experiment's <code>onSample</code> sends them, and aggregates exist only as a live fold over session logs.</p><h2 id="keep-assignments-durable" tabindex="-1">Keep assignments durable <a class="header-anchor" href="#keep-assignments-durable" aria-label="Permalink to "Keep assignments durable""></a></h2><p>The append-only <code>events.ndjson</code> stream is the source of truth. Each <code>ab.assigned</code> event persists a variant key or null skip. Built-in metrics come from the turn and tool events that follow it.</p><p>After a server restart or a parked session resumes, the live collector replays the stream to rebuild cumulative counters. Replay does not call <code>onSample</code> (or <code>persistSamples</code>) for historical turns. Only a new completed or failed turn emits another sample.</p><p>The snapshot API also replays <code>derive</code> across the full stream, so custom totals match the current extractor. Changing a derive function can change historical snapshot totals. Treat metric definitions as versioned experiment code.</p><h2 id="keep-eval-traffic-separate" tabindex="-1">Keep eval traffic separate <a class="header-anchor" href="#keep-eval-traffic-separate" aria-label="Permalink to "Keep eval traffic separate""></a></h2><p>Sessions created by <code>agentkit eval</code> and the playground Evals runner use <code>purpose: "eval"</code>. They skip A/B enrollment entirely:</p><ul><li>No split function runs.</li><li>No <code>ab.assigned</code> event is recorded.</li><li>No <code>onSample</code> callback fires.</li><li>The session is omitted from <code>GET /v1/abs</code>.</li></ul><p>Ordinary chat, <code>agentkit run</code>, Slack, GitHub, and other channel sessions use the live purpose. You do not need <code>splitIf</code> to exclude eval traffic.</p><h2 id="know-the-boundaries" tabindex="-1">Know the boundaries <a class="header-anchor" href="#know-the-boundaries" aria-label="Permalink to "Know the boundaries""></a></h2><p><code>defineAB</code> provides sticky assignment, variant instructions, <code>session.abs</code> for tools, cumulative metrics, and local inspection. It does not provide:</p><ul><li>A test command, assertion API, or pass/fail result</li><li>Statistical significance calculations</li><li>An experiment rollout or lifecycle service</li><li>Per-variant model or runtime configuration</li><li>A built-in analytics warehouse (bring your own via <code>persistSamples</code> / <code>persistSnapshots</code> / <code>onSample</code>)</li></ul><p>Use <a href="./evals.html">evals</a> to protect known behavior. Configure <code>agent/ab.config.ts</code> (or <code>onSample</code>) when you need sample/snapshot exports beyond the session event log.</p><h2 id="what-s-next" tabindex="-1">What's next <a class="header-anchor" href="#what-s-next" aria-label="Permalink to "What's next""></a></h2><p>Continue with these pages:</p><ul><li><a href="./evals.html">Evals</a>: pass/fail regression checks on fixed inputs</li><li><a href="./hillclimbing.html">Hillclimbing</a>: improve an agent against fixed fixtures</li><li><a href="./reference/hooks.html">Hooks</a>: other event-stream consumers</li><li><a href="./reference/sessions.html">Sessions and streaming</a>: the <code>ab.assigned</code> event and durable log</li><li><a href="./reference/playground.html">Playground</a>: the A/Bs tab</li><li><a href="./reference/http-api.html">HTTP API</a>: <code>GET /v1/abs</code></li><li><a href="./../skills/ab/SKILL.html">Live A/B metrics skill</a>: have a coding agent wire an experiment</li></ul></div></div></main><footer class="VPDocFooter" data-v-53b9fda6 data-v-3958cc3f><!--[--><!--]--><!----><nav class="prev-next" aria-labelledby="doc-footer-aria-label" data-v-3958cc3f><span class="visually-hidden" id="doc-footer-aria-label" data-v-3958cc3f>Pager</span><div class="pager" data-v-3958cc3f><a class="VPLink link pager-link prev" href="/docs/evals.html" data-v-3958cc3f><!--[--><span class="desc" data-v-3958cc3f>Previous page</span><span class="title" data-v-3958cc3f>Evals</span><!--]--></a></div><div class="pager" data-v-3958cc3f><a class="VPLink link pager-link next" href="/docs/hillclimbing.html" data-v-3958cc3f><!--[--><span class="desc" data-v-3958cc3f>Next page</span><span class="title" data-v-3958cc3f>Hillclimbing</span><!--]--></a></div></nav></footer><!--[--><!--]--></div></div></div><!--[--><!--]--></div></div><!----><!--[--><!--]--></div></div>
|
|
84
|
+
<script>window.__VP_HASH_MAP__=JSON.parse("{\"ab.md\":\"COdXkces\",\"building-with-agents.md\":\"CJCtZCyi\",\"concepts.md\":\"Cfb9b-k1\",\"deployment.md\":\"Dx1TYNk5\",\"evals.md\":\"DPZ_MAnI\",\"guides_agent-to-agent.md\":\"CrtrsySy\",\"guides_cloud-runtime.md\":\"CYlNMTNp\",\"guides_github.md\":\"DwbKhCeS\",\"guides_human-in-the-loop.md\":\"Dvuctx7s\",\"guides_slack.md\":\"bv41fHfW\",\"guides_webhooks.md\":\"hFTik3lf\",\"hillclimbing.md\":\"DHNast08\",\"index.md\":\"BPKcj5AI\",\"quickstart.md\":\"tVPiGK_L\",\"reference_agent-config.md\":\"Bpd7HQwf\",\"reference_channels.md\":\"D7JTR03W\",\"reference_cli.md\":\"DA730zCu\",\"reference_connections.md\":\"C3vNH_DE\",\"reference_hooks.md\":\"BCEc3MyM\",\"reference_http-api.md\":\"DBAahtdz\",\"reference_instructions.md\":\"BC05LEQ8\",\"reference_playground.md\":\"CPZhfYaO\",\"reference_project-layout.md\":\"Bd_CKtNS\",\"reference_schedules.md\":\"D7qijxLk\",\"reference_sessions.md\":\"DdjWabnN\",\"reference_skills.md\":\"VQnlBT3Q\",\"reference_subagents.md\":\"CIRAVcPK\",\"reference_tools.md\":\"DF5kwlt0\",\"scaffolding-agents.md\":\"CyYfWGdc\",\"troubleshooting.md\":\"DkovjS7h\"}");window.__VP_SITE_DATA__=JSON.parse("{\"lang\":\"en-US\",\"dir\":\"ltr\",\"title\":\"agentkit\",\"description\":\"Filesystem-first framework for building and serving Cursor agents.\",\"base\":\"/docs/\",\"head\":[],\"router\":{\"prefetchLinks\":true},\"appearance\":true,\"themeConfig\":{\"nav\":[{\"text\":\"Quickstart\",\"link\":\"/quickstart\"},{\"text\":\"Guides\",\"link\":\"/guides/webhooks\"},{\"text\":\"Reference\",\"link\":\"/reference/project-layout\"}],\"search\":{\"provider\":\"local\"},\"outline\":{\"level\":[2,3]},\"sidebar\":[{\"text\":\"Getting started\",\"items\":[{\"text\":\"Overview\",\"link\":\"/\"},{\"text\":\"Quickstart\",\"link\":\"/quickstart\"},{\"text\":\"Scaffold an agent with Cursor\",\"link\":\"/scaffolding-agents\"},{\"text\":\"Concepts\",\"link\":\"/concepts\"}]},{\"text\":\"Improving agents\",\"items\":[{\"text\":\"Building agents with agents\",\"link\":\"/building-with-agents\"},{\"text\":\"Evals\",\"link\":\"/evals\"},{\"text\":\"Live A/B metrics\",\"link\":\"/ab\"},{\"text\":\"Hillclimbing\",\"link\":\"/hillclimbing\"}]},{\"text\":\"Guides\",\"items\":[{\"text\":\"Webhooks & custom channels\",\"link\":\"/guides/webhooks\"},{\"text\":\"GitHub\",\"link\":\"/guides/github\"},{\"text\":\"Slack\",\"link\":\"/guides/slack\"},{\"text\":\"Human-in-the-loop approvals\",\"link\":\"/guides/human-in-the-loop\"},{\"text\":\"Agent-to-agent\",\"link\":\"/guides/agent-to-agent\"},{\"text\":\"Cloud runtime\",\"link\":\"/guides/cloud-runtime\"}]},{\"text\":\"Operating\",\"items\":[{\"text\":\"Deployment\",\"link\":\"/deployment\"},{\"text\":\"Fix common problems\",\"link\":\"/troubleshooting\"}]},{\"text\":\"Reference\",\"collapsed\":false,\"items\":[{\"text\":\"Project layout\",\"link\":\"/reference/project-layout\"},{\"text\":\"Agent config\",\"link\":\"/reference/agent-config\"},{\"text\":\"Instructions\",\"link\":\"/reference/instructions\"},{\"text\":\"Tools\",\"link\":\"/reference/tools\"},{\"text\":\"Skills\",\"link\":\"/reference/skills\"},{\"text\":\"MCP Connections\",\"link\":\"/reference/connections\"},{\"text\":\"Subagents\",\"link\":\"/reference/subagents\"},{\"text\":\"Channels\",\"link\":\"/reference/channels\"},{\"text\":\"Schedules & reminders\",\"link\":\"/reference/schedules\"},{\"text\":\"Hooks\",\"link\":\"/reference/hooks\"},{\"text\":\"Sessions & streaming\",\"link\":\"/reference/sessions\"},{\"text\":\"Playground\",\"link\":\"/reference/playground\"},{\"text\":\"CLI\",\"link\":\"/reference/cli\"},{\"text\":\"HTTP API\",\"link\":\"/reference/http-api\"}]}]},\"locales\":{},\"scrollOffset\":134,\"cleanUrls\":false}");</script>
|
|
85
|
+
|
|
86
|
+
</body>
|
|
87
|
+
</html>
|
|
@@ -0,0 +1,63 @@
|
|
|
1
|
+
import{_ as e,c as t,o as i,ag as a}from"./chunks/framework.CAZyNGu9.js";const k=JSON.parse('{"title":"Live A/B metrics","description":"Assign sticky variants with defineAB, collect live performance metrics, and inspect per-arm results.","frontmatter":{"title":"Live A/B metrics","description":"Assign sticky variants with defineAB, collect live performance metrics, and inspect per-arm results."},"headers":[],"relativePath":"ab.md","filePath":"ab.md"}'),n={name:"ab.md"};function o(l,s,r,d,p,h){return i(),t("div",null,[...s[0]||(s[0]=[a(`<h1 id="live-a-b-metrics" tabindex="-1">Live A/B metrics <a class="header-anchor" href="#live-a-b-metrics" aria-label="Permalink to "Live A/B metrics""></a></h1><p>Use <code>defineAB</code> to compare variants on live agent sessions. New sessions receive a sticky assignment in each enrolled experiment. agentkit folds their durable event streams into tool, token, failure, and wall-time metrics. You can send cumulative samples to your metrics backend and inspect aggregates in the playground.</p><p><code>defineAB</code> compares live variants through sticky assignment, instruction overlays, optional tool branches, and cumulative metrics. Metric callbacks observe the result without approving, rejecting, or failing a turn. Use <a href="./evals.html">evals</a> for pass/fail regression checks on fixed inputs.</p><div class="note custom-block github-alert"><p class="custom-block-title">NOTE</p><p>Import paths here use <code>@anysphere/agentkit/ab</code>. On projects still using <code>@anysphere/agent-serve</code>, swap the import. See <a href="./README.html#run-the-cli">Run the CLI</a> for the full rename table.</p></div><h2 id="choose-live-a-b-metrics-or-evals" tabindex="-1">Choose live A/B metrics or evals <a class="header-anchor" href="#choose-live-a-b-metrics-or-evals" aria-label="Permalink to "Choose live A/B metrics or evals""></a></h2><p>Both features read the session event stream, but they answer different questions.</p><table tabindex="0"><thead><tr><th></th><th>Live A/B metrics</th><th>Evals</th></tr></thead><tbody><tr><td>Question</td><td>How do variants compare on live sessions?</td><td>Does the agent still meet a fixed contract?</td></tr><tr><td>Location</td><td><code>agent/ab.ts</code> or <code>agent/ab/<name>.ts</code></td><td><code>evals/**/*.eval.ts</code></td></tr><tr><td>Input</td><td>Dev or production traffic</td><td>Frozen prompts and fixtures</td></tr><tr><td>Output</td><td>Cumulative metrics by session and arm</td><td>Pass/fail assertions</td></tr><tr><td>How it runs</td><td>Automatically on new live sessions</td><td><code>agentkit eval</code></td></tr></tbody></table><p>There is no <code>agentkit ab</code> command or assertion API.</p><h2 id="define-an-experiment" tabindex="-1">Define an experiment <a class="header-anchor" href="#define-an-experiment" aria-label="Permalink to "Define an experiment""></a></h2><p>Author one experiment in <code>agent/ab.ts</code>, add more under <code>agent/ab/<name>.ts</code>, or use both forms. Each file defines one experiment. The experiment name comes from <code>name</code> when set. Otherwise, agentkit uses <code>ab</code> for <code>agent/ab.ts</code> and the file stem for files under <code>agent/ab/</code>.</p><div class="language-ts vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">ts</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;">// agent/ab/concise-weather.ts</span></span>
|
|
2
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">import</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> {</span></span>
|
|
3
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> defineAB,</span></span>
|
|
4
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> splitBySessionHash,</span></span>
|
|
5
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">} </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">from</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "@anysphere/agentkit/ab"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">;</span></span>
|
|
6
|
+
<span class="line"></span>
|
|
7
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">export</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> default</span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> defineAB</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">({</span></span>
|
|
8
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> name: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"concise-weather"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
9
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> variants: {</span></span>
|
|
10
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> control: {</span></span>
|
|
11
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> label: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"Baseline"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
12
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
13
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> treatment: {</span></span>
|
|
14
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> label: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"Short replies"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
15
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> description: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"Adds a one-paragraph response limit."</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
16
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> instructions: </span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"Keep weather replies to one short paragraph."</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
17
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
18
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
19
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> split: </span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">splitBySessionHash</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">({</span></span>
|
|
20
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> weights: { control: </span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;">1</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">, treatment: </span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;">1</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
21
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> holdout: </span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;">0.1</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
22
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> }),</span></span>
|
|
23
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> derive: {</span></span>
|
|
24
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> weatherCalls</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">: (</span><span style="--shiki-light:#E36209;--shiki-dark:#FFAB70;">event</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">) </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">=></span></span>
|
|
25
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> event.type </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">===</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "action.result"</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> &&</span></span>
|
|
26
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> event.data.toolName </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">===</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "get_weather"</span></span>
|
|
27
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> ?</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> 1</span></span>
|
|
28
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> :</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> null</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
29
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
30
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> onSample</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span><span style="--shiki-light:#E36209;--shiki-dark:#FFAB70;">sample</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">) {</span></span>
|
|
31
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> console.</span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">log</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span></span>
|
|
32
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> sample.experiment,</span></span>
|
|
33
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> sample.variant,</span></span>
|
|
34
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> sample.metrics.toolCalls,</span></span>
|
|
35
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> sample.metrics.wallTimeMs</span></span>
|
|
36
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> );</span></span>
|
|
37
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> },</span></span>
|
|
38
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">});</span></span></code></pre></div><p>Every definition needs:</p><ul><li>At least two variants. Variant keys cannot be empty or contain <code>/</code> or <code>\\</code>.</li><li>A <code>split</code> function that returns a variant key or <code>null</code>.</li><li>An <code>onSample</code> callback for completed or failed turns.</li></ul><p><code>label</code> and <code>description</code> appear with the arm in result surfaces. <code>instructions</code> changes the prompt for sessions in that arm. <code>derive</code> adds custom counters.</p><p>Duplicate experiment names are validation errors. Check discovery before you serve:</p><div class="language-bash vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">bash</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> validate</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span></span>
|
|
39
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> info</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --json</span></span></code></pre></div><p>The <code>abs</code> field in <code>info</code> lists the discovered experiment names.</p><h2 id="assign-sticky-variants" tabindex="-1">Assign sticky variants <a class="header-anchor" href="#assign-sticky-variants" aria-label="Permalink to "Assign sticky variants""></a></h2><p>Enrollment happens once, when a live session is created and before its first turn:</p><ol><li>agentkit records <code>session.started</code>.</li><li>Each experiment runs its <code>split</code> function.</li><li>agentkit records one durable <code>ab.assigned</code> event per experiment.</li><li>The selected arms become available on <code>session.abs</code>.</li><li>Variant instruction overlays reach the first model turn.</li></ol><p>A split can return a variant key or <code>null</code>. A null assignment is a sticky skip for that experiment. It increments the experiment's <code>skipped</code> total, still appears in the snapshot's <code>sessions</code> list with <code>variant: null</code>, and does not collect arm metrics or call <code>onSample</code>.</p><p>Use the split helper that matches your rollout:</p><table tabindex="0"><thead><tr><th>Helper</th><th>Behavior</th></tr></thead><tbody><tr><td><code>splitBySessionHash({ weights?, holdout?, salt? })</code></td><td>Hashes the session id into a reproducible arm; the recommended default</td></tr><tr><td><code>splitByRandom({ weights?, holdout? })</code></td><td>Draws once when the session starts, then persists the result</td></tr><tr><td><code>splitAlways("control")</code></td><td>Pins every new session to one arm</td></tr><tr><td><code>splitNone()</code></td><td>Skips every new session without deleting the experiment</td></tr><tr><td><code>splitIf(predicate, inner)</code></td><td>Runs <code>inner</code> only when the predicate passes</td></tr><tr><td>Custom <code>split(ctx)</code></td><td>Returns a declared variant key or <code>null</code></td></tr></tbody></table><p>The split context includes the agent name, channel id, session info, experiment name, and declared variant keys. For example, enroll only Slack sessions:</p><div class="language-ts vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">ts</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">split</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">: </span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">splitIf</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span></span>
|
|
40
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> (</span><span style="--shiki-light:#E36209;--shiki-dark:#FFAB70;">ctx</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">) </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">=></span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> ctx.channel.id </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">===</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "slack"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">,</span></span>
|
|
41
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> splitBySessionHash</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">()</span></span>
|
|
42
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">),</span></span></code></pre></div><p>Weights default to equal. Non-positive weights leave an arm out of the draw, and at least one arm must have a positive weight. <code>holdout</code> is the fraction of sessions assigned <code>null</code>, from <code>0</code> through <code>1</code>. Change <code>salt</code> to reshuffle future hash assignments without renaming the experiment.</p><p>If a custom split throws or returns an unknown variant, agentkit logs the error and records <code>variant: null</code>. The failed decision becomes a sticky skip instead of breaking the session.</p><p>Enrollment only applies to new sessions. Adding an experiment does not assign existing conversations. Follow-ups keep the session's original arms. Keep experiment names and variant keys stable while you collect and compare results.</p><h2 id="change-behavior-by-variant" tabindex="-1">Change behavior by variant <a class="header-anchor" href="#change-behavior-by-variant" aria-label="Permalink to "Change behavior by variant""></a></h2><p>Variant instructions are appended to the agent's base instructions. Local sessions receive the merged instructions in <code>AGENTS.md</code> before every turn. Cloud sessions receive them in the first-turn preamble only. For cloud follow-ups, branch through <code>session.abs</code> when the arm must remain visible to deterministic behavior.</p><p>Tools can branch on the assignment through <code>ctx.session.abs</code>. Hooks can read the same field for logging or export:</p><div class="language-ts vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">ts</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">const</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> treatment</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> =</span></span>
|
|
43
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> ctx.session.abs?.[</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">"concise-weather"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">] </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">===</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "treatment"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">;</span></span>
|
|
44
|
+
<span class="line"></span>
|
|
45
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">if</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> (treatment) {</span></span>
|
|
46
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> return</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> conciseWeatherResult;</span></span>
|
|
47
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">}</span></span>
|
|
48
|
+
<span class="line"></span>
|
|
49
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">return</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> baselineWeatherResult;</span></span></code></pre></div><p>This makes the assignment available to deterministic code as well as the model prompt. Use both patterns together when one experiment must steer the prompt and host code at once.</p><p><code>defineAB</code> does not select a different model or runtime for each arm. Keep those settings in <code>agent/agent.ts</code>, or write explicit host logic when your experiment needs another behavior lever.</p><p>The split and selected arm can affect agent behavior. <code>derive</code> and <code>onSample</code> only observe the resulting event stream. Errors in either callback are logged and never fail the turn.</p><h2 id="collect-built-in-and-custom-metrics" tabindex="-1">Collect built-in and custom metrics <a class="header-anchor" href="#collect-built-in-and-custom-metrics" aria-label="Permalink to "Collect built-in and custom metrics""></a></h2><p>Metrics accumulate for each session and experiment. When one session joins several experiments, every enrolled experiment folds the same turn and tool events into its own counters.</p><table tabindex="0"><thead><tr><th>Metric</th><th>How agentkit calculates it</th></tr></thead><tbody><tr><td><code>turns</code></td><td>Adds one on <code>turn.completed</code> or <code>turn.failed</code></td></tr><tr><td><code>turnFailures</code></td><td>Adds one on <code>turn.failed</code></td></tr><tr><td><code>toolCalls</code></td><td>Adds one for each <code>action.result</code></td></tr><tr><td><code>toolErrors</code></td><td>Adds one when <code>action.result.data.isError</code> is true</td></tr><tr><td><code>inputTokens</code>, <code>outputTokens</code></td><td>Adds usage from completed turns</td></tr><tr><td><code>cacheReadTokens</code>, <code>cacheWriteTokens</code></td><td>Adds cache usage from completed turns</td></tr><tr><td><code>wallTimeMs</code></td><td>Sums the time from <code>turn.started</code> to its completed or failed event</td></tr><tr><td><code>custom</code></td><td>Sums finite numeric deltas returned by <code>derive</code></td></tr></tbody></table><p><code>onSample</code> fires after every <code>turn.completed</code> and <code>turn.failed</code> event for an enrolled arm. The sample contains:</p><table tabindex="0"><thead><tr><th>Field</th><th>Value</th></tr></thead><tbody><tr><td><code>experiment</code></td><td>Experiment name</td></tr><tr><td><code>variant</code>, <code>variantLabel?</code></td><td>Sticky arm and optional display label</td></tr><tr><td><code>sessionId</code>, <code>channelId</code></td><td>Source session</td></tr><tr><td><code>metrics</code></td><td>Cumulative metrics through this turn</td></tr><tr><td><code>reason</code></td><td><code>turn.completed</code> or <code>turn.failed</code></td></tr><tr><td><code>at</code></td><td>Terminal event timestamp</td></tr></tbody></table><p>The metrics are cumulative, not per-turn deltas. A second sample from the same session includes the first turn's counts.</p><p>Each <code>derive</code> extractor runs on every session event for its enrolled experiment, including streamed <code>message.appended</code> events. Keep it synchronous and cheap. Return a finite number to add a delta, or <code>null</code> to skip the event. Send samples to your metrics service from <code>onSample</code>; do not perform network or disk work in <code>derive</code>.</p><p>Skipped sessions never call <code>onSample</code>. Errors from <code>derive</code> or <code>onSample</code> are logged, then metric collection continues.</p><h2 id="inspect-assignments-and-results" tabindex="-1">Inspect assignments and results <a class="header-anchor" href="#inspect-assignments-and-results" aria-label="Permalink to "Inspect assignments and results""></a></h2><p>Open the playground's <strong>A/Bs</strong> tab to see aggregate arm totals and per-session assignments. The tab reads <code>GET /v1/abs</code>.</p><p>The response has two views of the same durable data:</p><table tabindex="0"><thead><tr><th>Field</th><th>Contents</th></tr></thead><tbody><tr><td><code>experiments</code></td><td>Declared variants, skipped-session count, arm session counts, and aggregate metrics</td></tr><tr><td><code>sessions</code></td><td>Visible sessions with their assignments and cumulative metrics</td></tr></tbody></table><p><code>GET /v1/abs</code> returns sessions visible to the current principal by default. In <code>--dev</code>, loopback requests include every session. Add <code>--allow-anonymous</code> to include every session from non-loopback callers too. This include-all behavior can still apply to <code>GET /v1/abs</code> in dev when bearer or custom auth keeps <code>GET /v1/sessions</code> owner-scoped.</p><p>Session <code>events.ndjson</code> is the source of truth for assignment + fold. <code>GET /v1/abs</code> recomputes aggregates from those logs. Use <code>agent/ab.config.ts</code> (and/or each experiment's <code>onSample</code>) when you want author-controlled sample/snapshot exports — same idea as eval <code>persistRuns</code>.</p><h2 id="configure-retention-and-persistence" tabindex="-1">Configure retention and persistence <a class="header-anchor" href="#configure-retention-and-persistence" aria-label="Permalink to "Configure retention and persistence""></a></h2><p>Optional project defaults in <code>agent/ab.config.ts</code>:</p><div class="language-ts vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">ts</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">import</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> {</span></span>
|
|
50
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> defineABConfig,</span></span>
|
|
51
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> persistABSamplesToDir,</span></span>
|
|
52
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> persistABSnapshotsToDir,</span></span>
|
|
53
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">} </span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">from</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "@anysphere/agentkit/ab"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">;</span></span>
|
|
54
|
+
<span class="line"></span>
|
|
55
|
+
<span class="line"><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">export</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"> default</span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;"> defineABConfig</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">({</span></span>
|
|
56
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // Optional — defaults to 200. Only affects GET /v1/abs / A/Bs tab.</span></span>
|
|
57
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // maxPlaygroundSessions: 200,</span></span>
|
|
58
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // retainSnapshots: 20,</span><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // prune budget for persistABSnapshotsToDir</span></span>
|
|
59
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // Append onSample payloads (in addition to each experiment's onSample)</span></span>
|
|
60
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> persistSamples: </span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">persistABSamplesToDir</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">".agent-serve/ab-samples"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">),</span></span>
|
|
61
|
+
<span class="line"><span style="--shiki-light:#6A737D;--shiki-dark:#6A737D;"> // Save aggregate snapshots whenever GET /v1/abs runs</span></span>
|
|
62
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;"> persistSnapshots: </span><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">persistABSnapshotsToDir</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">(</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">".agent-serve/ab-snapshots"</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">),</span></span>
|
|
63
|
+
<span class="line"><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">});</span></span></code></pre></div><table tabindex="0"><thead><tr><th>Option</th><th>Default</th><th>Meaning</th></tr></thead><tbody><tr><td><code>maxPlaygroundSessions</code></td><td><code>200</code></td><td>Max newest sessions folded into the playground / <code>GET /v1/abs</code> (not session logs or <code>persistSamples</code>)</td></tr><tr><td><code>persistSamples</code></td><td>unset</td><td>Durable sink for metric samples (<code>save(sample, ctx)</code>)</td></tr><tr><td><code>persistSnapshots</code></td><td>unset</td><td>Durable store for aggregate snapshots (<code>load</code> / <code>save</code> / optional <code>delete</code>)</td></tr><tr><td><code>retainSnapshots</code></td><td><code>20</code> when snapshots persist</td><td>Snapshot files kept after prune (<code>persistABSnapshotsToDir</code> honors this)</td></tr></tbody></table><p>Implement your own <code>{ save }</code> / <code>{ load, save, delete? }</code> for S3, a DB, or your metrics vendor. Without <code>persistSamples</code> / <code>persistSnapshots</code>, samples only go where each experiment's <code>onSample</code> sends them, and aggregates exist only as a live fold over session logs.</p><h2 id="keep-assignments-durable" tabindex="-1">Keep assignments durable <a class="header-anchor" href="#keep-assignments-durable" aria-label="Permalink to "Keep assignments durable""></a></h2><p>The append-only <code>events.ndjson</code> stream is the source of truth. Each <code>ab.assigned</code> event persists a variant key or null skip. Built-in metrics come from the turn and tool events that follow it.</p><p>After a server restart or a parked session resumes, the live collector replays the stream to rebuild cumulative counters. Replay does not call <code>onSample</code> (or <code>persistSamples</code>) for historical turns. Only a new completed or failed turn emits another sample.</p><p>The snapshot API also replays <code>derive</code> across the full stream, so custom totals match the current extractor. Changing a derive function can change historical snapshot totals. Treat metric definitions as versioned experiment code.</p><h2 id="keep-eval-traffic-separate" tabindex="-1">Keep eval traffic separate <a class="header-anchor" href="#keep-eval-traffic-separate" aria-label="Permalink to "Keep eval traffic separate""></a></h2><p>Sessions created by <code>agentkit eval</code> and the playground Evals runner use <code>purpose: "eval"</code>. They skip A/B enrollment entirely:</p><ul><li>No split function runs.</li><li>No <code>ab.assigned</code> event is recorded.</li><li>No <code>onSample</code> callback fires.</li><li>The session is omitted from <code>GET /v1/abs</code>.</li></ul><p>Ordinary chat, <code>agentkit run</code>, Slack, GitHub, and other channel sessions use the live purpose. You do not need <code>splitIf</code> to exclude eval traffic.</p><h2 id="know-the-boundaries" tabindex="-1">Know the boundaries <a class="header-anchor" href="#know-the-boundaries" aria-label="Permalink to "Know the boundaries""></a></h2><p><code>defineAB</code> provides sticky assignment, variant instructions, <code>session.abs</code> for tools, cumulative metrics, and local inspection. It does not provide:</p><ul><li>A test command, assertion API, or pass/fail result</li><li>Statistical significance calculations</li><li>An experiment rollout or lifecycle service</li><li>Per-variant model or runtime configuration</li><li>A built-in analytics warehouse (bring your own via <code>persistSamples</code> / <code>persistSnapshots</code> / <code>onSample</code>)</li></ul><p>Use <a href="./evals.html">evals</a> to protect known behavior. Configure <code>agent/ab.config.ts</code> (or <code>onSample</code>) when you need sample/snapshot exports beyond the session event log.</p><h2 id="what-s-next" tabindex="-1">What's next <a class="header-anchor" href="#what-s-next" aria-label="Permalink to "What's next""></a></h2><p>Continue with these pages:</p><ul><li><a href="./evals.html">Evals</a>: pass/fail regression checks on fixed inputs</li><li><a href="./hillclimbing.html">Hillclimbing</a>: improve an agent against fixed fixtures</li><li><a href="./reference/hooks.html">Hooks</a>: other event-stream consumers</li><li><a href="./reference/sessions.html">Sessions and streaming</a>: the <code>ab.assigned</code> event and durable log</li><li><a href="./reference/playground.html">Playground</a>: the A/Bs tab</li><li><a href="./reference/http-api.html">HTTP API</a>: <code>GET /v1/abs</code></li><li><a href="./../skills/ab/SKILL.html">Live A/B metrics skill</a>: have a coding agent wire an experiment</li></ul>`,69)])])}const u=e(n,[["render",o]]);export{k as __pageData,u as default};
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
import{_ as e,c as t,o as i,ag as a}from"./chunks/framework.CAZyNGu9.js";const k=JSON.parse('{"title":"Live A/B metrics","description":"Assign sticky variants with defineAB, collect live performance metrics, and inspect per-arm results.","frontmatter":{"title":"Live A/B metrics","description":"Assign sticky variants with defineAB, collect live performance metrics, and inspect per-arm results."},"headers":[],"relativePath":"ab.md","filePath":"ab.md"}'),n={name:"ab.md"};function o(l,s,r,d,p,h){return i(),t("div",null,[...s[0]||(s[0]=[a("",69)])])}const u=e(n,[["render",o]]);export{k as __pageData,u as default};
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
import{t as p}from"./chunks/theme.CZRvu_0q.js";import{R as s,a2 as i,a3 as u,a4 as c,a5 as l,a6 as f,a7 as d,a8 as m,a9 as h,aa as g,ab as A,d as v,u as y,v as C,s as P,ac as b,ad as w,ae as R,af as E}from"./chunks/framework.CAZyNGu9.js";function r(e){if(e.extends){const a=r(e.extends);return{...a,...e,async enhanceApp(t){a.enhanceApp&&await a.enhanceApp(t),e.enhanceApp&&await e.enhanceApp(t)}}}return e}const n=r(p),S=v({name:"VitePressApp",setup(){const{site:e,lang:a,dir:t}=y();return C(()=>{P(()=>{document.documentElement.lang=a.value,document.documentElement.dir=t.value})}),e.value.router.prefetchLinks&&b(),w(),R(),n.setup&&n.setup(),()=>E(n.Layout)}});async function T(){globalThis.__VITEPRESS__=!0;const e=_(),a=D();a.provide(u,e);const t=c(e.route);return a.provide(l,t),a.component("Content",f),a.component("ClientOnly",d),Object.defineProperties(a.config.globalProperties,{$frontmatter:{get(){return t.frontmatter.value}},$params:{get(){return t.page.value.params}}}),n.enhanceApp&&await n.enhanceApp({app:a,router:e,siteData:m}),{app:a,router:e,data:t}}function D(){return A(S)}function _(){let e=s;return h(a=>{let t=g(a),o=null;return t&&(e&&(t=t.replace(/\.js$/,".lean.js")),o=import(t)),s&&(e=!1),o},n.NotFound)}s&&T().then(({app:e,router:a,data:t})=>{a.go().then(()=>{i(a.route,t.site),e.mount("#app")})});export{T as createApp};
|
|
@@ -0,0 +1,13 @@
|
|
|
1
|
+
import{_ as a,c as t,o as i,ag as s}from"./chunks/framework.CAZyNGu9.js";const k=JSON.parse('{"title":"Building agents with agents","description":"Use the built-in skills to have a coding agent create, verify, and improve your agent.","frontmatter":{"title":"Building agents with agents","description":"Use the built-in skills to have a coding agent create, verify, and improve your agent."},"headers":[],"relativePath":"building-with-agents.md","filePath":"building-with-agents.md"}'),n={name:"building-with-agents.md"};function l(h,e,r,o,d,c){return i(),t("div",null,[...e[0]||(e[0]=[s(`<h1 id="building-agents-with-agents" tabindex="-1">Building agents with agents <a class="header-anchor" href="#building-agents-with-agents" aria-label="Permalink to "Building agents with agents""></a></h1><p>Give a coding agent the goal. The built-in skills guide it through scaffolding, channels, verification, evals, and measured improvement.</p><h2 id="what-can-a-coding-agent-build-for-me" tabindex="-1">What can a coding agent build for me? <a class="header-anchor" href="#what-can-a-coding-agent-build-for-me" aria-label="Permalink to "What can a coding agent build for me?""></a></h2><p>A coding agent can take a agentkit project from an idea to a verified first version. It can:</p><ul><li>Create the project and choose the smallest useful surface</li><li>Add tools, skills, channels, MCP connections, and approval gates</li><li>Run a real turn and inspect its trace</li><li>Add evals for the behavior you want to keep</li><li>Improve the agent against fixed inputs</li></ul><p>This works because the project surface lives in files. The CLI also returns structured JSON and records each run as an NDJSON trace. Your coding agent can inspect what the framework discovered, run the agent, and verify the result without reading terminal prose.</p><h2 id="how-do-i-create-an-agent-with-the-built-in-skill" tabindex="-1">How do I create an agent with the built-in skill? <a class="header-anchor" href="#how-do-i-create-an-agent-with-the-built-in-skill" aria-label="Permalink to "How do I create an agent with the built-in skill?""></a></h2><p>Have the coding agent read <a href="./../skills/create-agent/SKILL.html"><code>skills/create-agent/SKILL.md</code></a> (in the monorepo: <code>packages/agent-serve/skills/create-agent/SKILL.md</code>) and follow it.</p><p>The skill asks about your agent's purpose, runtime, model, channels, MCP connections, and capabilities. It then shows you a plan, writes the project, and verifies the result.</p><p>For example:</p><blockquote><p>Use the agent-serve create-agent skill to build a PR triage agent reachable through GitHub. It should summarize failed checks, require approval before posting a review, and include one smoke eval.</p></blockquote><p>The skill starts with a small surface: short instructions, one or two tools, the channels you chose, and a smoke eval. You can add more after the first end-to-end turn works.</p><h2 id="which-built-in-skill-should-i-use" tabindex="-1">Which built-in skill should I use? <a class="header-anchor" href="#which-built-in-skill-should-i-use" aria-label="Permalink to "Which built-in skill should I use?""></a></h2><p>The package ships task-specific guides under <a href="./../skills/"><code>skills/</code></a>:</p><table tabindex="0"><thead><tr><th>What you want to do</th><th>Skill</th></tr></thead><tbody><tr><td>Understand the project layout and runtimes</td><td><a href="./../skills/framework-map/SKILL.html"><code>framework-map</code></a></td></tr><tr><td>Create and verify a new agent</td><td><a href="./../skills/create-agent/SKILL.html"><code>create-agent</code></a></td></tr><tr><td>Write fixtures and regression checks</td><td><a href="./../skills/evals/SKILL.html"><code>evals</code></a></td></tr><tr><td>Live A/B metrics on traffic (<code>defineAB</code>)</td><td><a href="./../skills/ab/SKILL.html"><code>ab</code></a></td></tr><tr><td>Improve an agent against fixed inputs</td><td><a href="./../skills/hillclimb/SKILL.html"><code>hillclimb</code></a></td></tr><tr><td>Add GitHub webhooks and replay events</td><td><a href="./../skills/github/SKILL.html"><code>github</code></a></td></tr><tr><td>Connect an agent to Slack</td><td><a href="./../skills/setup-slack/SKILL.html"><code>setup-slack</code></a></td></tr><tr><td>Diagnose a local run</td><td><a href="./../skills/debug/SKILL.html"><code>debug</code></a></td></tr></tbody></table><p>Point your coding agent at the matching <code>SKILL.md</code>. The guide contains the workflow, commands, and common mistakes for that task.</p><div class="note custom-block github-alert"><p class="custom-block-title">NOTE</p><p>The skill bodies use the current <code>agent-serve</code> CLI names. This guide uses the upcoming <code>agentkit</code> names. See <a href="./README.html#run-the-cli">Run the CLI</a> for the full rename table.</p></div><h2 id="how-does-a-coding-agent-verify-its-work" tabindex="-1">How does a coding agent verify its work? <a class="header-anchor" href="#how-does-a-coding-agent-verify-its-work" aria-label="Permalink to "How does a coding agent verify its work?""></a></h2><p>The coding agent should discover the project, test each server tool, run a real turn, and finish with evals:</p><div class="language-bash vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">bash</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> validate</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span></span>
|
|
2
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> info</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --json</span></span>
|
|
3
|
+
<span class="line"></span>
|
|
4
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> call</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> inspect_pr</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> \\</span></span>
|
|
5
|
+
<span class="line"><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --input</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> '{"prUrl":"https://github.com/acme/checkout/pull/42"}'</span></span>
|
|
6
|
+
<span class="line"></span>
|
|
7
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> run</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> \\</span></span>
|
|
8
|
+
<span class="line"><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --message</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> "Is https://github.com/acme/checkout/pull/42 ready to approve?"</span></span>
|
|
9
|
+
<span class="line"></span>
|
|
10
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> trajectory</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --events</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .agentkit/traces/</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;"><</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">sessionI</span><span style="--shiki-light:#24292E;--shiki-dark:#E1E4E8;">d</span><span style="--shiki-light:#D73A49;--shiki-dark:#F97583;">></span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;">.ndjson</span></span>
|
|
11
|
+
<span class="line"></span>
|
|
12
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> eval</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --list</span></span>
|
|
13
|
+
<span class="line"><span style="--shiki-light:#6F42C1;--shiki-dark:#B392F0;">agentkit</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> eval</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --dir</span><span style="--shiki-light:#032F62;--shiki-dark:#9ECBFF;"> .</span><span style="--shiki-light:#005CC5;--shiki-dark:#79B8FF;"> --json</span></span></code></pre></div><p><code>serve</code> boots without an API key, which is enough to check channel mounts and the playground shell. Model turns need a credential. When the environment lacks one, finish every key-free check, then hand the <code>run</code> and <code>eval</code> commands to the operator.</p><p>Test server tools with <code>call</code> before tuning the prompt. It runs a tool in-process with schema validation and no model turn. If the tool returns the wrong data, a prompt change won't fix it.</p><p><code>validate</code> and <code>run</code> don't type-check the project because tsx strips types. Run the project's TypeScript check before shipping. Tool results must also be JSON-shaped. Use object literals or <code>type</code> aliases for <code>execute</code> return types instead of <code>interface</code> types.</p><h2 id="how-do-i-improve-the-agent-after-its-first-run" tabindex="-1">How do I improve the agent after its first run? <a class="header-anchor" href="#how-do-i-improve-the-agent-after-its-first-run" aria-label="Permalink to "How do I improve the agent after its first run?""></a></h2><p>Scaffolding proves the agent runs. It says nothing about quality. Once a smoke turn passes, give the hillclimb skill:</p><ol><li><strong>Fixtures</strong>: one to three fixed inputs, such as a PR URL, a canonical question, or a saved webhook payload</li><li><strong>Success criteria</strong>: correct tool choice, fewer tool calls, lower wall time, or better output</li><li><strong>The freeze line</strong>: the API contract, output shape, and existing evals that must stay unchanged</li></ol><p>Have the coding agent read <a href="./../skills/hillclimb/SKILL.html"><code>skills/hillclimb/SKILL.md</code></a>. It measures the current run, proposes one change, remeasures the same fixtures, and adds an eval for each kept improvement.</p><h2 id="related" tabindex="-1">Related <a class="header-anchor" href="#related" aria-label="Permalink to "Related""></a></h2><ul><li><a href="./quickstart.html">Create your first agent</a></li><li><a href="./scaffolding-agents.html">Scaffold an agent with Cursor</a></li><li><a href="./evals.html">Evals</a></li><li><a href="./hillclimbing.html">Hillclimbing</a></li><li><a href="./reference/cli.html">CLI reference</a></li></ul>`,29)])])}const g=a(n,[["render",l]]);export{k as __pageData,g as default};
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
import{_ as a,c as t,o as i,ag as s}from"./chunks/framework.CAZyNGu9.js";const k=JSON.parse('{"title":"Building agents with agents","description":"Use the built-in skills to have a coding agent create, verify, and improve your agent.","frontmatter":{"title":"Building agents with agents","description":"Use the built-in skills to have a coding agent create, verify, and improve your agent."},"headers":[],"relativePath":"building-with-agents.md","filePath":"building-with-agents.md"}'),n={name:"building-with-agents.md"};function l(h,e,r,o,d,c){return i(),t("div",null,[...e[0]||(e[0]=[s("",29)])])}const g=a(n,[["render",l]]);export{k as __pageData,g as default};
|