@apralabs/apra-fleet 0.4.0 → 0.4.1
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/README.md +145 -7
- package/dist/agents/backlog-groomer.md +17 -1
- package/dist/agents/ci-watcher.md +13 -1
- package/dist/agents/deployer.md +15 -3
- package/dist/agents/doer.md +35 -1
- package/dist/agents/harvester.md +22 -3
- package/dist/agents/integ-test-runner.md +18 -2
- package/dist/agents/kb-reconciler.md +216 -0
- package/dist/agents/plan-reviewer.md +14 -1
- package/dist/agents/planner.md +88 -5
- package/dist/agents/regression-test-runner.md +19 -1
- package/dist/agents/reviewer.md +104 -3
- package/dist/agents/schemas/doer-output.json +63 -3
- package/dist/agents/schemas/harvester-output.json +59 -2
- package/dist/agents/schemas/integ-test-runner-output.json +2 -2
- package/dist/agents/schemas/kb-reconciler-output.json +87 -0
- package/dist/agents/schemas/planner-output.json +65 -0
- package/dist/agents/schemas/reviewer-output.json +105 -8
- package/dist/cli/install.d.ts +1 -0
- package/dist/cli/install.d.ts.map +1 -1
- package/dist/cli/install.js +136 -5
- package/dist/cli/install.js.map +1 -1
- package/dist/cli/join.js +1 -1
- package/dist/cli/join.js.map +1 -1
- package/dist/cli/kb-commit.d.ts +12 -0
- package/dist/cli/kb-commit.d.ts.map +1 -0
- package/dist/cli/kb-commit.js +53 -0
- package/dist/cli/kb-commit.js.map +1 -0
- package/dist/cli/kb-directives.d.ts +17 -0
- package/dist/cli/kb-directives.d.ts.map +1 -0
- package/dist/cli/kb-directives.js +146 -0
- package/dist/cli/kb-directives.js.map +1 -0
- package/dist/cli/kb-import.d.ts +13 -0
- package/dist/cli/kb-import.d.ts.map +1 -0
- package/dist/cli/kb-import.js +56 -0
- package/dist/cli/kb-import.js.map +1 -0
- package/dist/cli/uninstall.d.ts.map +1 -1
- package/dist/cli/uninstall.js +11 -0
- package/dist/cli/uninstall.js.map +1 -1
- package/dist/cli/watch.d.ts +53 -4
- package/dist/cli/watch.d.ts.map +1 -1
- package/dist/cli/watch.js +118 -17
- package/dist/cli/watch.js.map +1 -1
- package/dist/commands/kb-server.d.ts +8 -0
- package/dist/commands/kb-server.d.ts.map +1 -0
- package/dist/commands/kb-server.js +221 -0
- package/dist/commands/kb-server.js.map +1 -0
- package/dist/index.js +77 -2
- package/dist/index.js.map +1 -1
- package/dist/os/linux.d.ts +1 -0
- package/dist/os/linux.d.ts.map +1 -1
- package/dist/os/linux.js +18 -1
- package/dist/os/linux.js.map +1 -1
- package/dist/os/os-commands.d.ts +5 -0
- package/dist/os/os-commands.d.ts.map +1 -1
- package/dist/os/windows.d.ts +35 -0
- package/dist/os/windows.d.ts.map +1 -1
- package/dist/os/windows.js +67 -31
- package/dist/os/windows.js.map +1 -1
- package/dist/providers/agy.d.ts +12 -1
- package/dist/providers/agy.d.ts.map +1 -1
- package/dist/providers/agy.js +134 -9
- package/dist/providers/agy.js.map +1 -1
- package/dist/providers/claude.d.ts +11 -1
- package/dist/providers/claude.d.ts.map +1 -1
- package/dist/providers/claude.js +47 -19
- package/dist/providers/claude.js.map +1 -1
- package/dist/providers/codex.d.ts +11 -1
- package/dist/providers/codex.d.ts.map +1 -1
- package/dist/providers/codex.js +29 -6
- package/dist/providers/codex.js.map +1 -1
- package/dist/providers/copilot.d.ts +11 -1
- package/dist/providers/copilot.d.ts.map +1 -1
- package/dist/providers/copilot.js +26 -6
- package/dist/providers/copilot.js.map +1 -1
- package/dist/providers/email/headers.d.ts +15 -0
- package/dist/providers/email/headers.d.ts.map +1 -0
- package/dist/providers/email/headers.js +42 -0
- package/dist/providers/email/headers.js.map +1 -0
- package/dist/providers/email/index.d.ts +18 -0
- package/dist/providers/email/index.d.ts.map +1 -0
- package/dist/providers/email/index.js +60 -0
- package/dist/providers/email/index.js.map +1 -0
- package/dist/providers/email/provider.d.ts +49 -0
- package/dist/providers/email/provider.d.ts.map +1 -0
- package/dist/providers/email/provider.js +2 -0
- package/dist/providers/email/provider.js.map +1 -0
- package/dist/providers/email/sendgrid.d.ts +12 -0
- package/dist/providers/email/sendgrid.d.ts.map +1 -0
- package/dist/providers/email/sendgrid.js +67 -0
- package/dist/providers/email/sendgrid.js.map +1 -0
- package/dist/providers/email/smtp.d.ts +65 -0
- package/dist/providers/email/smtp.d.ts.map +1 -0
- package/dist/providers/email/smtp.js +373 -0
- package/dist/providers/email/smtp.js.map +1 -0
- package/dist/providers/gemini.d.ts +11 -1
- package/dist/providers/gemini.d.ts.map +1 -1
- package/dist/providers/gemini.js +38 -4
- package/dist/providers/gemini.js.map +1 -1
- package/dist/providers/none.d.ts +11 -1
- package/dist/providers/none.d.ts.map +1 -1
- package/dist/providers/none.js +21 -0
- package/dist/providers/none.js.map +1 -1
- package/dist/providers/opencode.d.ts +11 -1
- package/dist/providers/opencode.d.ts.map +1 -1
- package/dist/providers/opencode.js +55 -4
- package/dist/providers/opencode.js.map +1 -1
- package/dist/providers/provider.d.ts +77 -0
- package/dist/providers/provider.d.ts.map +1 -1
- package/dist/providers/provider.js +44 -0
- package/dist/providers/provider.js.map +1 -1
- package/dist/services/agent-provisioner.d.ts.map +1 -1
- package/dist/services/agent-provisioner.js +3 -6
- package/dist/services/agent-provisioner.js.map +1 -1
- package/dist/services/cloud/lifecycle.d.ts.map +1 -1
- package/dist/services/cloud/lifecycle.js +3 -0
- package/dist/services/cloud/lifecycle.js.map +1 -1
- package/dist/services/cloud/task-wrapper.d.ts +18 -0
- package/dist/services/cloud/task-wrapper.d.ts.map +1 -1
- package/dist/services/cloud/task-wrapper.js +157 -0
- package/dist/services/cloud/task-wrapper.js.map +1 -1
- package/dist/services/github-app.d.ts.map +1 -1
- package/dist/services/github-app.js +11 -2
- package/dist/services/github-app.js.map +1 -1
- package/dist/services/knowledge/audn.d.ts +37 -0
- package/dist/services/knowledge/audn.d.ts.map +1 -0
- package/dist/services/knowledge/audn.js +219 -0
- package/dist/services/knowledge/audn.js.map +1 -0
- package/dist/services/knowledge/file-hash.d.ts +11 -0
- package/dist/services/knowledge/file-hash.d.ts.map +1 -0
- package/dist/services/knowledge/file-hash.js +123 -0
- package/dist/services/knowledge/file-hash.js.map +1 -0
- package/dist/services/knowledge/http-provider.d.ts +46 -0
- package/dist/services/knowledge/http-provider.d.ts.map +1 -0
- package/dist/services/knowledge/http-provider.js +254 -0
- package/dist/services/knowledge/http-provider.js.map +1 -0
- package/dist/services/knowledge/kb-providers.d.ts +15 -0
- package/dist/services/knowledge/kb-providers.d.ts.map +1 -0
- package/dist/services/knowledge/kb-providers.js +92 -0
- package/dist/services/knowledge/kb-providers.js.map +1 -0
- package/dist/services/knowledge/kb-scope-input.d.ts +5 -0
- package/dist/services/knowledge/kb-scope-input.d.ts.map +1 -0
- package/dist/services/knowledge/kb-scope-input.js +11 -0
- package/dist/services/knowledge/kb-scope-input.js.map +1 -0
- package/dist/services/knowledge/kb-service.d.ts +3 -0
- package/dist/services/knowledge/kb-service.d.ts.map +1 -0
- package/dist/services/knowledge/kb-service.js +2 -0
- package/dist/services/knowledge/kb-service.js.map +1 -0
- package/dist/services/knowledge/path-validation.d.ts +2 -0
- package/dist/services/knowledge/path-validation.d.ts.map +1 -0
- package/dist/services/knowledge/path-validation.js +13 -0
- package/dist/services/knowledge/path-validation.js.map +1 -0
- package/dist/services/knowledge/pre-init.d.ts +14 -0
- package/dist/services/knowledge/pre-init.d.ts.map +1 -0
- package/dist/services/knowledge/pre-init.js +147 -0
- package/dist/services/knowledge/pre-init.js.map +1 -0
- package/dist/services/knowledge/project-slug.d.ts +2 -0
- package/dist/services/knowledge/project-slug.d.ts.map +1 -0
- package/dist/services/knowledge/project-slug.js +49 -0
- package/dist/services/knowledge/project-slug.js.map +1 -0
- package/dist/services/knowledge/sqlite-provider.d.ts +147 -0
- package/dist/services/knowledge/sqlite-provider.d.ts.map +1 -0
- package/dist/services/knowledge/sqlite-provider.js +1646 -0
- package/dist/services/knowledge/sqlite-provider.js.map +1 -0
- package/dist/services/knowledge/types.d.ts +163 -0
- package/dist/services/knowledge/types.d.ts.map +1 -0
- package/dist/services/knowledge/types.js +9 -0
- package/dist/services/knowledge/types.js.map +1 -0
- package/dist/services/member-home.d.ts +89 -0
- package/dist/services/member-home.d.ts.map +1 -0
- package/dist/services/member-home.js +256 -0
- package/dist/services/member-home.js.map +1 -0
- package/dist/services/member-remote-url.d.ts +34 -0
- package/dist/services/member-remote-url.d.ts.map +1 -0
- package/dist/services/member-remote-url.js +39 -0
- package/dist/services/member-remote-url.js.map +1 -0
- package/dist/services/orphan-recovery.d.ts +5 -2
- package/dist/services/orphan-recovery.d.ts.map +1 -1
- package/dist/services/orphan-recovery.js +19 -8
- package/dist/services/orphan-recovery.js.map +1 -1
- package/dist/services/preflight-check.d.ts +56 -0
- package/dist/services/preflight-check.d.ts.map +1 -0
- package/dist/services/preflight-check.js +277 -0
- package/dist/services/preflight-check.js.map +1 -0
- package/dist/services/stall/log-path-resolver.d.ts +11 -8
- package/dist/services/stall/log-path-resolver.d.ts.map +1 -1
- package/dist/services/stall/log-path-resolver.js +17 -40
- package/dist/services/stall/log-path-resolver.js.map +1 -1
- package/dist/services/stall/stall-detector.d.ts +32 -0
- package/dist/services/stall/stall-detector.d.ts.map +1 -1
- package/dist/services/stall/stall-detector.js +149 -6
- package/dist/services/stall/stall-detector.js.map +1 -1
- package/dist/services/stall/stall-poller.d.ts +34 -0
- package/dist/services/stall/stall-poller.d.ts.map +1 -1
- package/dist/services/stall/stall-poller.js +134 -2
- package/dist/services/stall/stall-poller.js.map +1 -1
- package/dist/services/strategy.d.ts.map +1 -1
- package/dist/services/strategy.js +2 -2
- package/dist/services/strategy.js.map +1 -1
- package/dist/services/tool-registry.d.ts.map +1 -1
- package/dist/services/tool-registry.js +87 -1
- package/dist/services/tool-registry.js.map +1 -1
- package/dist/services/vcs/github.d.ts.map +1 -1
- package/dist/services/vcs/github.js +32 -1
- package/dist/services/vcs/github.js.map +1 -1
- package/dist/services/version-check.d.ts +22 -0
- package/dist/services/version-check.d.ts.map +1 -0
- package/dist/services/version-check.js +103 -0
- package/dist/services/version-check.js.map +1 -0
- package/dist/skills/fleet-sprint-cli/SKILL.md +13 -2
- package/dist/skills/pm/beads.md +6 -0
- package/dist/skills/pm/fleet-addendum.md +6 -0
- package/dist/skills/pm/index.md +56 -0
- package/dist/skills/pm/kb-reconcile.md +132 -0
- package/dist/skills/pm/kb-review.md +109 -0
- package/dist/tools/check-status.d.ts +47 -0
- package/dist/tools/check-status.d.ts.map +1 -1
- package/dist/tools/check-status.js +242 -0
- package/dist/tools/check-status.js.map +1 -1
- package/dist/tools/child-id-allocator.d.ts +2 -2
- package/dist/tools/code-intelligence-codebase-memory.d.ts +11 -0
- package/dist/tools/code-intelligence-codebase-memory.d.ts.map +1 -0
- package/dist/tools/code-intelligence-codebase-memory.js +270 -0
- package/dist/tools/code-intelligence-codebase-memory.js.map +1 -0
- package/dist/tools/code-intelligence-freshness.d.ts +2 -0
- package/dist/tools/code-intelligence-freshness.d.ts.map +1 -0
- package/dist/tools/code-intelligence-freshness.js +26 -0
- package/dist/tools/code-intelligence-freshness.js.map +1 -0
- package/dist/tools/code-intelligence-gitnexus.d.ts +14 -0
- package/dist/tools/code-intelligence-gitnexus.d.ts.map +1 -0
- package/dist/tools/code-intelligence-gitnexus.js +475 -0
- package/dist/tools/code-intelligence-gitnexus.js.map +1 -0
- package/dist/tools/code-intelligence-kb-enrich.d.ts +2 -0
- package/dist/tools/code-intelligence-kb-enrich.d.ts.map +1 -0
- package/dist/tools/code-intelligence-kb-enrich.js +65 -0
- package/dist/tools/code-intelligence-kb-enrich.js.map +1 -0
- package/dist/tools/code-intelligence-reindex.d.ts +7 -0
- package/dist/tools/code-intelligence-reindex.d.ts.map +1 -0
- package/dist/tools/code-intelligence-reindex.js +106 -0
- package/dist/tools/code-intelligence-reindex.js.map +1 -0
- package/dist/tools/code-intelligence-telemetry.d.ts +5 -0
- package/dist/tools/code-intelligence-telemetry.d.ts.map +1 -0
- package/dist/tools/code-intelligence-telemetry.js +52 -0
- package/dist/tools/code-intelligence-telemetry.js.map +1 -0
- package/dist/tools/code-intelligence-tests.d.ts +2 -0
- package/dist/tools/code-intelligence-tests.d.ts.map +1 -0
- package/dist/tools/code-intelligence-tests.js +24 -0
- package/dist/tools/code-intelligence-tests.js.map +1 -0
- package/dist/tools/code-intelligence.d.ts +114 -0
- package/dist/tools/code-intelligence.d.ts.map +1 -0
- package/dist/tools/code-intelligence.js +124 -0
- package/dist/tools/code-intelligence.js.map +1 -0
- package/dist/tools/compose-permissions.d.ts +26 -0
- package/dist/tools/compose-permissions.d.ts.map +1 -1
- package/dist/tools/compose-permissions.js +244 -33
- package/dist/tools/compose-permissions.js.map +1 -1
- package/dist/tools/execute-command.d.ts +2 -0
- package/dist/tools/execute-command.d.ts.map +1 -1
- package/dist/tools/execute-command.js +86 -16
- package/dist/tools/execute-command.js.map +1 -1
- package/dist/tools/execute-prompt.d.ts +13 -1
- package/dist/tools/execute-prompt.d.ts.map +1 -1
- package/dist/tools/execute-prompt.js +257 -45
- package/dist/tools/execute-prompt.js.map +1 -1
- package/dist/tools/kb-capture.d.ts +53 -0
- package/dist/tools/kb-capture.d.ts.map +1 -0
- package/dist/tools/kb-capture.js +130 -0
- package/dist/tools/kb-capture.js.map +1 -0
- package/dist/tools/kb-context.d.ts +17 -0
- package/dist/tools/kb-context.d.ts.map +1 -0
- package/dist/tools/kb-context.js +29 -0
- package/dist/tools/kb-context.js.map +1 -0
- package/dist/tools/kb-export.d.ts +19 -0
- package/dist/tools/kb-export.d.ts.map +1 -0
- package/dist/tools/kb-export.js +344 -0
- package/dist/tools/kb-export.js.map +1 -0
- package/dist/tools/kb-feedback.d.ts +23 -0
- package/dist/tools/kb-feedback.d.ts.map +1 -0
- package/dist/tools/kb-feedback.js +43 -0
- package/dist/tools/kb-feedback.js.map +1 -0
- package/dist/tools/kb-freshness-sweep.d.ts +14 -0
- package/dist/tools/kb-freshness-sweep.d.ts.map +1 -0
- package/dist/tools/kb-freshness-sweep.js +29 -0
- package/dist/tools/kb-freshness-sweep.js.map +1 -0
- package/dist/tools/kb-harvest.d.ts +20 -0
- package/dist/tools/kb-harvest.d.ts.map +1 -0
- package/dist/tools/kb-harvest.js +143 -0
- package/dist/tools/kb-harvest.js.map +1 -0
- package/dist/tools/kb-import.d.ts +44 -0
- package/dist/tools/kb-import.d.ts.map +1 -0
- package/dist/tools/kb-import.js +239 -0
- package/dist/tools/kb-import.js.map +1 -0
- package/dist/tools/kb-invalidate.d.ts +19 -0
- package/dist/tools/kb-invalidate.d.ts.map +1 -0
- package/dist/tools/kb-invalidate.js +34 -0
- package/dist/tools/kb-invalidate.js.map +1 -0
- package/dist/tools/kb-list.d.ts +32 -0
- package/dist/tools/kb-list.d.ts.map +1 -0
- package/dist/tools/kb-list.js +43 -0
- package/dist/tools/kb-list.js.map +1 -0
- package/dist/tools/kb-promote.d.ts +20 -0
- package/dist/tools/kb-promote.d.ts.map +1 -0
- package/dist/tools/kb-promote.js +20 -0
- package/dist/tools/kb-promote.js.map +1 -0
- package/dist/tools/kb-query.d.ts +35 -0
- package/dist/tools/kb-query.d.ts.map +1 -0
- package/dist/tools/kb-query.js +111 -0
- package/dist/tools/kb-query.js.map +1 -0
- package/dist/tools/kb-reconcile-prefilter.d.ts +14 -0
- package/dist/tools/kb-reconcile-prefilter.d.ts.map +1 -0
- package/dist/tools/kb-reconcile-prefilter.js +26 -0
- package/dist/tools/kb-reconcile-prefilter.js.map +1 -0
- package/dist/tools/kb-resolve-contradiction.d.ts +23 -0
- package/dist/tools/kb-resolve-contradiction.d.ts.map +1 -0
- package/dist/tools/kb-resolve-contradiction.js +37 -0
- package/dist/tools/kb-resolve-contradiction.js.map +1 -0
- package/dist/tools/kb-session-prime.d.ts +26 -0
- package/dist/tools/kb-session-prime.d.ts.map +1 -0
- package/dist/tools/kb-session-prime.js +392 -0
- package/dist/tools/kb-session-prime.js.map +1 -0
- package/dist/tools/kb-setup.d.ts +20 -0
- package/dist/tools/kb-setup.d.ts.map +1 -0
- package/dist/tools/kb-setup.js +49 -0
- package/dist/tools/kb-setup.js.map +1 -0
- package/dist/tools/kb-stats.d.ts +20 -0
- package/dist/tools/kb-stats.d.ts.map +1 -0
- package/dist/tools/kb-stats.js +115 -0
- package/dist/tools/kb-stats.js.map +1 -0
- package/dist/tools/member-detail.d.ts.map +1 -1
- package/dist/tools/member-detail.js +8 -0
- package/dist/tools/member-detail.js.map +1 -1
- package/dist/tools/monitor-task.d.ts.map +1 -1
- package/dist/tools/monitor-task.js +21 -3
- package/dist/tools/monitor-task.js.map +1 -1
- package/dist/tools/provision-auth.d.ts.map +1 -1
- package/dist/tools/provision-auth.js +10 -3
- package/dist/tools/provision-auth.js.map +1 -1
- package/dist/tools/register-member.d.ts +3 -0
- package/dist/tools/register-member.d.ts.map +1 -1
- package/dist/tools/register-member.js +30 -1
- package/dist/tools/register-member.js.map +1 -1
- package/dist/tools/remove-member.d.ts.map +1 -1
- package/dist/tools/remove-member.js +17 -3
- package/dist/tools/remove-member.js.map +1 -1
- package/dist/tools/report-status.d.ts.map +1 -1
- package/dist/tools/report-status.js +2 -1
- package/dist/tools/report-status.js.map +1 -1
- package/dist/tools/send-email.d.ts +69 -0
- package/dist/tools/send-email.d.ts.map +1 -0
- package/dist/tools/send-email.js +177 -0
- package/dist/tools/send-email.js.map +1 -0
- package/dist/tools/setup-ssh-key.d.ts.map +1 -1
- package/dist/tools/setup-ssh-key.js +2 -0
- package/dist/tools/setup-ssh-key.js.map +1 -1
- package/dist/tools/update-member.d.ts +3 -0
- package/dist/tools/update-member.d.ts.map +1 -1
- package/dist/tools/update-member.js +25 -1
- package/dist/tools/update-member.js.map +1 -1
- package/dist/types.d.ts +1 -0
- package/dist/types.d.ts.map +1 -1
- package/dist/utils/prompt-errors.js +1 -1
- package/dist/utils/prompt-errors.js.map +1 -1
- package/dist/utils/session-caller.d.ts +28 -0
- package/dist/utils/session-caller.d.ts.map +1 -0
- package/dist/utils/session-caller.js +24 -0
- package/dist/utils/session-caller.js.map +1 -0
- package/dist/utils/work-folder-validation.d.ts +39 -0
- package/dist/utils/work-folder-validation.d.ts.map +1 -0
- package/dist/utils/work-folder-validation.js +60 -0
- package/dist/utils/work-folder-validation.js.map +1 -0
- package/dist/workflows/auto-sprint.js +546 -6
- package/package.json +7 -4
- package/packages/apra-fleet-client/src/client/api.mjs +118 -4
- package/packages/apra-fleet-se/apra-pm/agents/backlog-groomer.md +17 -1
- package/packages/apra-fleet-se/apra-pm/agents/ci-watcher.md +13 -1
- package/packages/apra-fleet-se/apra-pm/agents/deployer.md +15 -3
- package/packages/apra-fleet-se/apra-pm/agents/doer.md +35 -1
- package/packages/apra-fleet-se/apra-pm/agents/harvester.md +22 -3
- package/packages/apra-fleet-se/apra-pm/agents/integ-test-runner.md +18 -2
- package/packages/apra-fleet-se/apra-pm/agents/kb-reconciler.md +216 -0
- package/packages/apra-fleet-se/apra-pm/agents/plan-reviewer.md +14 -1
- package/packages/apra-fleet-se/apra-pm/agents/planner.md +88 -5
- package/packages/apra-fleet-se/apra-pm/agents/regression-test-runner.md +19 -1
- package/packages/apra-fleet-se/apra-pm/agents/reviewer.md +104 -3
- package/packages/apra-fleet-se/apra-pm/agents/schemas/doer-output.json +63 -3
- package/packages/apra-fleet-se/apra-pm/agents/schemas/harvester-output.json +59 -2
- package/packages/apra-fleet-se/apra-pm/agents/schemas/integ-test-runner-output.json +2 -2
- package/packages/apra-fleet-se/apra-pm/agents/schemas/kb-reconciler-output.json +87 -0
- package/packages/apra-fleet-se/apra-pm/agents/schemas/planner-output.json +65 -0
- package/packages/apra-fleet-se/apra-pm/agents/schemas/reviewer-output.json +105 -8
- package/packages/apra-fleet-se/apra-pm/skills/pm/beads.md +6 -0
- package/packages/apra-fleet-se/apra-pm/skills/pm/fleet-addendum.md +6 -0
- package/packages/apra-fleet-se/apra-pm/skills/pm/index.md +56 -0
- package/packages/apra-fleet-se/apra-pm/skills/pm/kb-reconcile.md +132 -0
- package/packages/apra-fleet-se/apra-pm/skills/pm/kb-review.md +109 -0
- package/packages/apra-fleet-se/bin/cli.mjs +7 -0
- package/packages/apra-fleet-se/bin/serve.mjs +14 -2
- package/packages/apra-fleet-se/fleet-sprint/contracts.mjs +31 -5
- package/packages/apra-fleet-se/fleet-sprint/docs/README.md +1 -1
- package/packages/apra-fleet-se/fleet-sprint/docs/coach-role-design.md +23 -151
- package/packages/apra-fleet-se/fleet-sprint/docs/dolt-manual-recovery-verified.md +256 -0
- package/packages/apra-fleet-se/fleet-sprint/docs/dolt-sync-redesign.md +1482 -0
- package/packages/apra-fleet-se/fleet-sprint/docs/escalate-to-llm-design.md +908 -0
- package/packages/apra-fleet-se/fleet-sprint/docs/shared-orchestrator-reservation-design.md +399 -0
- package/packages/apra-fleet-se/fleet-sprint/dolt-settle.mjs +972 -0
- package/packages/apra-fleet-se/fleet-sprint/dolt-sync.mjs +167 -101
- package/packages/apra-fleet-se/fleet-sprint/errors.mjs +50 -6
- package/packages/apra-fleet-se/fleet-sprint/runner.js +1416 -116
- package/packages/apra-fleet-se/fleet-sprint/skills/fleet-sprint-cli/SKILL.md +13 -2
- package/packages/apra-fleet-se/fleet-sprint/skills/fleet-supervisor/SKILL.md +350 -0
- package/packages/apra-fleet-se/fleet-sprint/sprint-progress.mjs +101 -0
- package/packages/apra-fleet-se/fleet-sprint/vcs-providers/github.mjs +30 -17
- package/packages/apra-fleet-se/fleet-sprint/viewer-extensions.mjs +62 -2
- package/packages/apra-fleet-se/package.json +1 -0
- package/packages/apra-fleet-workflow/src/viewer/index.mjs +127 -13
- package/packages/apra-fleet-workflow/src/viewer/run-title.mjs +60 -0
- package/packages/apra-fleet-workflow/src/workflow/index.mjs +8 -4
- package/skills/fleet/knowledge-agent.md +66 -0
- package/skills/fleet/permissions.md +28 -4
- package/skills/fleet/profiles/base-dev.json +10 -1
- package/skills/fleet/profiles/base-reviewer.json +8 -1
- package/skills/fleet/profiles/tag-kb-reconciler.json +20 -0
- package/version.json +1 -1
- package/packages/apra-fleet-se/fleet-sprint/docs/dolt-tier2-runbook.md +0 -127
- package/packages/apra-fleet-se/fleet-sprint/dolt-recovery-path-b.mjs +0 -339
- package/packages/apra-fleet-se/fleet-sprint/dolt-recovery-tier2.mjs +0 -360
- package/packages/apra-fleet-se/fleet-sprint/dolt-recovery.mjs +0 -363
|
@@ -0,0 +1,109 @@
|
|
|
1
|
+
# /pm kb-review
|
|
2
|
+
|
|
3
|
+
Surface all contradiction-flagged KB entries and resolve them interactively.
|
|
4
|
+
|
|
5
|
+
Note: `/pm kb-review` also surfaces pending `user-directive` proposals (via
|
|
6
|
+
`kb_query({ flagged_only: true })`) and instructs the human to run
|
|
7
|
+
`apra-fleet kb approve-directive <id>` or `reject-directive <id>` -- these are
|
|
8
|
+
CLI-only resolutions, never `kb_promote`.
|
|
9
|
+
|
|
10
|
+
## When to run
|
|
11
|
+
|
|
12
|
+
- After any sprint completes, as part of KB hygiene.
|
|
13
|
+
- When an agent reports a `flagged` audn_decision from `kb_capture`.
|
|
14
|
+
- Periodically, to keep the KB free of unresolved contradictions.
|
|
15
|
+
|
|
16
|
+
## Steps
|
|
17
|
+
|
|
18
|
+
### Step 1: Fetch flagged entries
|
|
19
|
+
|
|
20
|
+
Call `kb_query(flagged_only=true)`.
|
|
21
|
+
|
|
22
|
+
- If the response says "No flagged contradictions found -- KB is clean.", report that to the user and stop.
|
|
23
|
+
- Otherwise, collect all returned entries.
|
|
24
|
+
|
|
25
|
+
### Step 2: Group into contradiction pairs
|
|
26
|
+
|
|
27
|
+
For each entry with `contradiction_of` set, pair it with the entry whose `id` matches `contradiction_of`.
|
|
28
|
+
Entries with `flagged_for_review=true` and no `contradiction_of` are the original (challenged) entries.
|
|
29
|
+
|
|
30
|
+
Group: `{ original: entry_A, challenger: entry_B }` where `entry_B.contradiction_of = entry_A.id`.
|
|
31
|
+
|
|
32
|
+
### Step 3: Present each pair to the user
|
|
33
|
+
|
|
34
|
+
For each pair, show:
|
|
35
|
+
|
|
36
|
+
```
|
|
37
|
+
Contradiction pair (N of M):
|
|
38
|
+
|
|
39
|
+
[ORIGINAL] id: <id> confidence: <confidence> created: <created_at>
|
|
40
|
+
Title: <title>
|
|
41
|
+
Summary: <summary>
|
|
42
|
+
Content: <content>
|
|
43
|
+
|
|
44
|
+
[CHALLENGER] id: <id> confidence: <confidence> created: <created_at>
|
|
45
|
+
Title: <title>
|
|
46
|
+
Summary: <summary>
|
|
47
|
+
Content: <content>
|
|
48
|
+
|
|
49
|
+
Resolution options:
|
|
50
|
+
A - Keep original, discard challenger
|
|
51
|
+
B - Keep challenger, discard original
|
|
52
|
+
M - Merge both into a new entry (you provide merged content)
|
|
53
|
+
D - Delete both (neither is correct)
|
|
54
|
+
S - Skip (defer to later)
|
|
55
|
+
```
|
|
56
|
+
|
|
57
|
+
Wait for the user's response before proceeding to the next pair.
|
|
58
|
+
|
|
59
|
+
### Step 4: Apply resolution
|
|
60
|
+
|
|
61
|
+
Supersede is OPT-IN: you must name the entry you are retiring via
|
|
62
|
+
`supersedes=<id>`. It is honored only when AUDN independently matches that same
|
|
63
|
+
entry (same type, overlapping symbols and source_files), so the corrective
|
|
64
|
+
capture must still carry the SAME title/symbols/source_files as its target --
|
|
65
|
+
otherwise nothing is retired and you get a second live entry instead. A capture
|
|
66
|
+
without `supersedes` never retires anything.
|
|
67
|
+
|
|
68
|
+
**A -- Keep original:**
|
|
69
|
+
- `kb_promote(original.id, reason="contradiction resolved: original kept")` -- upgrades its confidence.
|
|
70
|
+
- `kb_capture` a new entry with the SAME title/symbols/source_files as the challenger, corrected content that carries no contradiction keyword or polarity word (or AUDN will flag it again instead of updating), `confidence=UNVERIFIED`, and `supersedes=<challenger.id>`. That retires the challenger.
|
|
71
|
+
|
|
72
|
+
**B -- Keep challenger:**
|
|
73
|
+
- `kb_promote(challenger.id, reason="contradiction resolved: challenger kept")`.
|
|
74
|
+
- `kb_capture` a corrected entry with the SAME title/symbols/source_files as the original, corrected content free of contradiction/polarity words, and `supersedes=<original.id>`. That retires the original.
|
|
75
|
+
|
|
76
|
+
**M -- Merge:**
|
|
77
|
+
- Ask the user to provide the merged content.
|
|
78
|
+
- `kb_capture` a new entry with merged content and `supersedes=<id of whichever entry it replaces>`. Repeat for the second entry if both must be retired -- `supersedes` names ONE entry, so retiring both takes two captures.
|
|
79
|
+
- `kb_promote` the newly captured entry id.
|
|
80
|
+
|
|
81
|
+
**D -- Delete both:**
|
|
82
|
+
- `kb_capture` a retraction entry: `title=<original title>`, `content="Retracted: neither entry describes current behavior."` (no contradiction keyword or polarity word, or AUDN will flag it again instead of updating), `confidence=UNVERIFIED`, `supersedes=<original.id>`.
|
|
83
|
+
- `kb_capture` same for the challenger, with `supersedes=<challenger.id>`.
|
|
84
|
+
|
|
85
|
+
**S -- Skip:**
|
|
86
|
+
- No action. Note it in the session summary.
|
|
87
|
+
|
|
88
|
+
**Verified actual behavior (T3.7, F11 e2e proof) -- read this before reporting "resolved":**
|
|
89
|
+
- Superseding an entry (via the corrective `kb_capture` above with `supersedes` set, AUDN decision `update`) sets its `superseded_at` and `stale=1`. That entry then drops out of every future `kb_query({flagged_only:true})` listing -- the tool always excludes superseded entries.
|
|
90
|
+
- `kb_promote` does **NOT** clear `flagged_for_review` and does **NOT** clear `contradiction_of`. This means the entry you KEPT (promoted, not superseded) stays visible in `kb_query({flagged_only:true})` forever -- it is not automatically delisted just because it was resolved. Do not expect the pair to fully disappear from the flagged list; expect only the superseded side to disappear. Recognize an already-resolved pair by its counterpart's absence (or by checking `promoted_at`/`confidence` on the id you kept), not by the pair vanishing entirely.
|
|
91
|
+
- A `kb_feedback`-downvoted entry (`stale=1` + `flagged_for_review=1`, never superseded) stays listed under `flagged_only` until it is separately resolved the same way -- superseded via a corrective capture, or otherwise retired. Simply promoting a *different* entry never clears it.
|
|
92
|
+
|
|
93
|
+
### Step 5: Report
|
|
94
|
+
|
|
95
|
+
After all pairs are processed, report:
|
|
96
|
+
|
|
97
|
+
```
|
|
98
|
+
KB Review complete.
|
|
99
|
+
- Pairs resolved: N
|
|
100
|
+
- Pairs skipped: M
|
|
101
|
+
- Total flagged entries cleared: K
|
|
102
|
+
```
|
|
103
|
+
|
|
104
|
+
"Cleared" here means the SUPERSEDED side of each resolved pair (the one whose
|
|
105
|
+
`superseded_at` got set) -- per the verified behavior in Step 4, the kept/
|
|
106
|
+
promoted side of a resolved pair remains in `kb_query({flagged_only:true})`
|
|
107
|
+
indefinitely, so K is normally N, not 2N. Do not report a pair as fully gone
|
|
108
|
+
from the flagged list; report it as resolved (one side superseded, the other
|
|
109
|
+
promoted).
|
|
@@ -690,6 +690,13 @@ async function main() {
|
|
|
690
690
|
// explicitly rather than picking up the new core defaults.
|
|
691
691
|
stateSnapshotDir: 'sprint-logs',
|
|
692
692
|
stateSnapshotPrefix: 'sprint_',
|
|
693
|
+
// apra-fleet-dm5.2: publish the fields buildRunTitle() (apra-fleet-
|
|
694
|
+
// dm5.1) reads off state.args so the viewer header shows
|
|
695
|
+
// "<members> working <targetIssues> (<goal>)" instead of just the
|
|
696
|
+
// workflow name for a fleet-sprint run. Purely additive -- the core
|
|
697
|
+
// viewer's state.args stays optional/null for any other workflow
|
|
698
|
+
// that doesn't pass launchArgs, and no existing state key changes.
|
|
699
|
+
launchArgs: { members: validMembers, targetIssues, goal },
|
|
693
700
|
});
|
|
694
701
|
|
|
695
702
|
// apra-fleet-unw2.16, N14 (e): the viewer port was hardcoded with no
|
|
@@ -46,7 +46,8 @@ import { createBacklog, registerBacklogRoutes } from '../src/supervisor/backlog.
|
|
|
46
46
|
import { createDashboard, registerDashboardRoutes } from '../src/supervisor/dashboard.mjs';
|
|
47
47
|
import { createSprintController, registerSprintRoutes, defaultMemberOverlapGuard, ApiError } from '../src/supervisor/api.mjs';
|
|
48
48
|
import { createScopeGuard, formatScopeConflict } from '../src/supervisor/scope-overlap.mjs';
|
|
49
|
-
import { listFleetMembers } from '../src/supervisor/fleet-members.mjs';
|
|
49
|
+
import { listFleetMembers, executeFleetCommand } from '../src/supervisor/fleet-members.mjs';
|
|
50
|
+
import { createDoltOrphanSweep } from '../src/supervisor/dolt-orphan-sweep.mjs';
|
|
50
51
|
import { resolveFleetServerConnection } from './cli.mjs';
|
|
51
52
|
|
|
52
53
|
const SERVE_USAGE = `
|
|
@@ -252,7 +253,18 @@ export async function serveMain(argv = process.argv.slice(2)) {
|
|
|
252
253
|
// above for why no separate launch-form seam is constructed here).
|
|
253
254
|
const dashboard = createDashboard({ ledger, watchdog, backlog });
|
|
254
255
|
|
|
255
|
-
|
|
256
|
+
// docs/dolt-sync-redesign.md Part 3.3: kill any orphaned ephemeral
|
|
257
|
+
// `dolt sql-server` a mid-settle orchestrator death left behind on a
|
|
258
|
+
// member (settle's own finally covers every other path). Both
|
|
259
|
+
// collaborators use the same short-lived-MCP-connection pattern as
|
|
260
|
+
// listFleetMembers -- the supervisor never holds a standing fleet
|
|
261
|
+
// transport.
|
|
262
|
+
const doltOrphanSweep = createDoltOrphanSweep({
|
|
263
|
+
listMembers: () => listFleetMembers({ resolveConnection: resolveFleetServerConnection }),
|
|
264
|
+
execCommand: ({ member, command }) => executeFleetCommand({ member, command, resolveConnection: resolveFleetServerConnection }),
|
|
265
|
+
});
|
|
266
|
+
|
|
267
|
+
const supervisor = createSupervisor({ port, ledger, spawner, watchdog, dashboard, idAllocator, doltMutex, doltOrphanSweep });
|
|
256
268
|
registerIdAllocatorRoutes(supervisor, idAllocator, { readJsonBody, sendJson });
|
|
257
269
|
registerDoltMutexRoutes(supervisor, doltMutex, { readJsonBody, sendJson });
|
|
258
270
|
|
|
@@ -265,11 +265,18 @@ export function assertVersionPin(role, schema, expectedMajor) {
|
|
|
265
265
|
// signal for this second case.
|
|
266
266
|
//
|
|
267
267
|
// Roles allow-listed here are exempt because they legitimately have no
|
|
268
|
-
// output schema file, by design (not by omission)
|
|
269
|
-
//
|
|
270
|
-
//
|
|
271
|
-
//
|
|
272
|
-
|
|
268
|
+
// output schema file, by design (not by omission).
|
|
269
|
+
//
|
|
270
|
+
// EMPTY as of 2026-08-03. "planner" used to be listed here on the layering
|
|
271
|
+
// proposal's reasoning that the planner's output IS the beads DAG it creates,
|
|
272
|
+
// not a structured verdict object. The KB trust pipeline supersedes that: its
|
|
273
|
+
// Capture-roles decision accepts kb_captures from doer, reviewer, planner and
|
|
274
|
+
// harvester, and a role can only return a field the engine reads if it has an
|
|
275
|
+
// output contract to carry it. agents/schemas/planner-output.json now exists,
|
|
276
|
+
// so the planner is no longer schema-less and a MISSING planner-output.json is
|
|
277
|
+
// once again a real defect worth warning about.
|
|
278
|
+
// See docs/superpowers/specs/2026-08-03-kb-trust-pipeline-design.md, Phase 2.
|
|
279
|
+
export const ROLES_WITHOUT_OUTPUT_SCHEMA = Object.freeze(new Set([]));
|
|
273
280
|
|
|
274
281
|
/**
|
|
275
282
|
* Emits a loud console.warn when `resolveSchemasDir()`'s resolved directory
|
|
@@ -614,6 +621,25 @@ export const finalVerdict = {
|
|
|
614
621
|
required: ['id', 'reason'],
|
|
615
622
|
},
|
|
616
623
|
},
|
|
624
|
+
// apra-fleet-nx7: the Final Review may promote KB entries, exactly as a
|
|
625
|
+
// per-round review can. Without this field the final reviewer -- the one
|
|
626
|
+
// role that has read the whole diff and run the full suite, i.e. the
|
|
627
|
+
// strongest evidence any promoter ever holds -- had no way to return a
|
|
628
|
+
// promotion decision, so knowledge captured in a sprint's last round was
|
|
629
|
+
// stranded at INFERRED with nobody left to verify it. Same {id, reason}
|
|
630
|
+
// shape as reviewer-output.json's kb_promotions, executed by the same
|
|
631
|
+
// kbWork.apply path. Optional: promoting nothing is the common answer.
|
|
632
|
+
kb_promotions: {
|
|
633
|
+
type: 'array',
|
|
634
|
+
items: {
|
|
635
|
+
type: 'object',
|
|
636
|
+
properties: {
|
|
637
|
+
id: { type: 'string' },
|
|
638
|
+
reason: { type: 'string' },
|
|
639
|
+
},
|
|
640
|
+
required: ['id', 'reason'],
|
|
641
|
+
},
|
|
642
|
+
},
|
|
617
643
|
},
|
|
618
644
|
required: ['verdict', 'notes'],
|
|
619
645
|
};
|
|
@@ -90,7 +90,7 @@ Watch progress there rather than tailing raw stdout.
|
|
|
90
90
|
| `--role-map <json\|@file>` | | no | -- | JSON object mapping role -> member[], e.g. `'{"doer":["m1","m2"]}'`. Inline JSON or `@path/to/file.json`. |
|
|
91
91
|
| `--viewer-port <port>` | | no | `8080` | Port for the local dashboard viewer. |
|
|
92
92
|
| `--budget <usd>` | | no | unlimited | USD ceiling for this run's total estimated spend. |
|
|
93
|
-
| `--dispatch-timeout-s <s>` | | no | `
|
|
93
|
+
| `--dispatch-timeout-s <s>` | | no | `9000` | Per-dispatch time budget in seconds, applied as both the inactivity timeout and the hard ceiling on every agent dispatch (integ-test dispatch ceiling is 2x, regression-test ceiling is 3x). Minimum 60. Lower it for small sprints so a hung dispatch costs minutes, not 2.5 hours. |
|
|
94
94
|
| `--sync` | | no | off | Synced topology mode (orchestrator-bracketed git sync): members may sit on differing HEADs but must share the same origin URL and pass a `bd dolt pull` probe. Omitted uses shared-workspace mode (all members on the same HEAD). |
|
|
95
95
|
| `--help` | `-h` | | | Show the engine's help. |
|
|
96
96
|
|
|
@@ -1,151 +1,23 @@
|
|
|
1
|
-
# Coach Role
|
|
2
|
-
|
|
3
|
-
Status:
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
The error-handling hierarchy is three layers, in order:
|
|
25
|
-
|
|
26
|
-
1. **Typed handlers and ladders (deterministic, free).** errors.mjs typed
|
|
27
|
-
errors, turn-exhaustion resume ladders, Tier-2 conflict resolution,
|
|
28
|
-
busy-lock retries. The coach NEVER preempts these.
|
|
29
|
-
2. **Coach (JIT judgment, tokens).** Invoked only when layer 1 has no
|
|
30
|
-
match or its handling failed -- the unforeseen complement.
|
|
31
|
-
3. **Fail loudly (existing behavior).** If the coach cannot resolve (or
|
|
32
|
-
its per-class/per-run caps are hit), the normal failure path proceeds
|
|
33
|
-
unchanged. The coach never suppresses a failure it could not fix.
|
|
34
|
-
|
|
35
|
-
Red-state interception points (initial taxonomy, extend as discovered):
|
|
36
|
-
- Streak failure after doer dispatch/resume ladder exhaustion
|
|
37
|
-
- Dispatch errors with unrecognized reason (api_error, empty_response,
|
|
38
|
-
transport failures) after existing retries
|
|
39
|
-
- Sync bracket failures (G-push/G-pull/D-push/D-pull) after Tier-2
|
|
40
|
-
- Phase-level failures: deploy failed, integ runner crashed, harvester
|
|
41
|
-
failed
|
|
42
|
-
- Reviewer/planner contract violations after their built-in retry
|
|
43
|
-
- Stall detection firing; state corruption detected; member unreachable;
|
|
44
|
-
auth expiry mid-run
|
|
45
|
-
- Workflow-level abort paths (as evidence collector even when
|
|
46
|
-
unresolvable)
|
|
47
|
-
|
|
48
|
-
## Guardrails: coach.md carries them ALL (REFINED)
|
|
49
|
-
|
|
50
|
-
The contract file (agents/coach.md, upstream in apra-pm) is the single
|
|
51
|
-
home of every guardrail -- not scattered through engine code -- so the
|
|
52
|
-
guardrails ship with the role definition, version together, and appear in
|
|
53
|
-
every dispatch. Engine-side caps are enforcement backstops of the same
|
|
54
|
-
rules, never the only statement of them.
|
|
55
|
-
|
|
56
|
-
Guardrail set (all load-bearing):
|
|
57
|
-
- ACTION ALLOWLIST only: stash/commit WIP to a rescue branch, clear a
|
|
58
|
-
leaked lock, retry a push, restart a dead session, re-run an idempotent
|
|
59
|
-
step. Hard prohibitions: never force-push, never delete branches/files,
|
|
60
|
-
never mutate bead verdicts or close beads, never edit code content,
|
|
61
|
-
never act outside the member's working tree, never install software.
|
|
62
|
-
- SINGLE-SHOT: one intervention attempt per red event; then report and
|
|
63
|
-
fall back. The coach is never coached (no recursion).
|
|
64
|
-
- SAME-CLASS-TWICE POLICY: an error class (by fingerprint) may be coached
|
|
65
|
-
at most N times per run (N=2 initial); after that it must fail loudly
|
|
66
|
-
and demand an engine fix. Healing is a bridge, never a home.
|
|
67
|
-
- EVERY INTERVENTION FILES A REPORT -- no silent saves. Interventions are
|
|
68
|
-
first-class events in sprint state and appear in final-review evidence
|
|
69
|
-
(a PASS with 12 coach saves reads differently from a clean PASS).
|
|
70
|
-
- TARGET-AGNOSTIC: the coach knows the fleet/workflow/auto-sprint STACK,
|
|
71
|
-
never the target repo's specifics (product-vs-dogfood rule).
|
|
72
|
-
- TOKEN DISCIPLINE: invoked only at layer 2 (rare by construction);
|
|
73
|
-
per-run invocation cap; premium tier acceptable given rarity.
|
|
74
|
-
- REPRODUCIBILITY: interventions recorded in replayable form; mock/golden
|
|
75
|
-
tests stub coach outcomes deterministically.
|
|
76
|
-
|
|
77
|
-
## Bug reports and consent (REFINED): sanitize, then ask -- or standing consent
|
|
78
|
-
|
|
79
|
-
The coach always produces a SANITIZED, ANONYMIZED report locally. Whether
|
|
80
|
-
it leaves the machine is a separate, consent-gated step:
|
|
81
|
-
|
|
82
|
-
- Modes (fleet config, e.g. coach.telemetry): "never" | "ask" (default) |
|
|
83
|
-
"always".
|
|
84
|
-
- **ask (default)**: report saved under sprint-logs/coach-reports/; the
|
|
85
|
-
user is asked to consent AFTER THE FACT -- at sprint end (PR body /
|
|
86
|
-
dashboard / CLI summary), with the exact sanitized text shown. The
|
|
87
|
-
zero-auth mechanism: a pre-filled GitHub new-issue URL
|
|
88
|
-
(github.com/Apra-Labs/apra-fleet/issues/new?title=...&body=...) -- one
|
|
89
|
-
click, human-in-the-loop by construction, nothing uploads silently,
|
|
90
|
-
no token handling at all.
|
|
91
|
-
- **always (standing consent, mainly internal/trusting users)**: approved
|
|
92
|
-
once and for all via explicit config; reports auto-filed via gh CLI
|
|
93
|
-
using the user's own gh auth, or a scoped token stored in the fleet
|
|
94
|
-
credential store (referenced as a {{secure.*}} handle, never plaintext).
|
|
95
|
-
Standing consent is per-install, revocable, and recorded in the report
|
|
96
|
-
footer ("filed under standing consent").
|
|
97
|
-
- **never**: local reports only; nothing asks, nothing uploads.
|
|
98
|
-
|
|
99
|
-
Sanitization spec (applies BEFORE the report is even saved locally):
|
|
100
|
-
- Absolute paths -> stable placeholders (<work-folder>/..., <home>/...)
|
|
101
|
-
- Hostnames, usernames, IPs, emails -> redacted tokens
|
|
102
|
-
- Anything matching secret patterns (keys, tokens, {{secure.*}} values,
|
|
103
|
-
env dumps) -> stripped outright, never placeholdered
|
|
104
|
-
- Target-repo identifiers (repo name, branch names, bead titles) ->
|
|
105
|
-
optional redaction tier: internal users may keep them; the default
|
|
106
|
-
anonymizes to <target-repo>, <sprint-branch>
|
|
107
|
-
- Code snippets: not included by default; error text is truncated to the
|
|
108
|
-
minimal reproducing lines
|
|
109
|
-
- A stable ANONYMOUS install id (random UUID, no derivation from machine
|
|
110
|
-
identity) enables cross-report dedup without identifying the user
|
|
111
|
-
|
|
112
|
-
Report content (schema sketch):
|
|
113
|
-
- fingerprint: normalized error-class hash (for dedup and the
|
|
114
|
-
same-class-twice cap)
|
|
115
|
-
- context: engine version, OS, provider, phase, role being dispatched
|
|
116
|
-
- narrative: what failed, diagnosis, actionsTaken[], resolved: bool,
|
|
117
|
-
time lost
|
|
118
|
-
- issueDraft: { title, redactedBody, labels: [coach-report, <class>] }
|
|
119
|
-
- Dedup behavior upstream: same fingerprint appends an occurrence
|
|
120
|
-
comment to the existing issue rather than opening a duplicate.
|
|
121
|
-
|
|
122
|
-
## Implementation sketch (unchanged mechanics, wider trigger)
|
|
123
|
-
|
|
124
|
-
- Contract: agents/coach.md upstream in apra-pm (UPSTREAM DEPENDENCY like
|
|
125
|
-
eft.11: needs an apra-pm PR + submodule pointer bump). Inputs: error
|
|
126
|
-
text, phase, member, recent command history, sprint-state excerpt.
|
|
127
|
-
Tools: Read + tightly-scoped Bash. Output: the report schema above.
|
|
128
|
-
- Engine: a single coachOnRedState() boundary called from every red-state
|
|
129
|
-
interception point listed above (one choke point, not N copies), after
|
|
130
|
-
layer-1 handling. Interventions ledger in sprint state; caps enforced
|
|
131
|
-
engine-side as backstop; telemetry mode read from fleet config.
|
|
132
|
-
- Relation to the outside stabilization loop: coach handles in-run JIT
|
|
133
|
-
recovery; the outside loop remains for engine-level fixes; coach
|
|
134
|
-
reports feed both.
|
|
135
|
-
|
|
136
|
-
## Open design questions (to resolve before operationalizing)
|
|
137
|
-
|
|
138
|
-
1. Consent surface for "ask" mode: sprint-end CLI summary, PR body
|
|
139
|
-
section, dashboard card, or all three? (Recommend: PR body section +
|
|
140
|
-
dashboard card; the PR is where humans already look.)
|
|
141
|
-
2. Does the coach run on the ORCHESTRATOR member or the affected member?
|
|
142
|
-
(Recommend: orchestrator-side dispatch that may execute allowlisted
|
|
143
|
-
commands ON the affected member -- keeps one contract, avoids
|
|
144
|
-
provisioning coach agents everywhere.)
|
|
145
|
-
3. N values: per-class cap (initial 2) and per-run cap (initial 5) --
|
|
146
|
-
validate against real red-state frequency from run logs.
|
|
147
|
-
4. Should coach reports also feed a LOCAL knowledge base consulted on
|
|
148
|
-
later invocations within the same install (self-priming), or is that
|
|
149
|
-
scope creep for v1? (Bias: v2.)
|
|
150
|
-
5. Fingerprint algorithm: normalize paths/ids out of the error text, hash
|
|
151
|
-
the remainder -- needs a worked spec with collision examples.
|
|
1
|
+
# Coach Role (superseded -- see escalate-to-llm-design.md)
|
|
2
|
+
|
|
3
|
+
Status: SUPERSEDED 2026-08-10 by the sprint-doctor design in
|
|
4
|
+
[`escalate-to-llm-design.md`](escalate-to-llm-design.md). Kept as a pointer so
|
|
5
|
+
links and history resolve; the full original text is in git history
|
|
6
|
+
(this file, prior to this commit).
|
|
7
|
+
|
|
8
|
+
The coach idea (2026-07-20): an in-workflow LLM role for error classes the
|
|
9
|
+
deterministic engine did not foresee -- invoked strictly after typed
|
|
10
|
+
handlers/ladders, single-shot, capped per error class, filing a sanitized
|
|
11
|
+
bug report for every intervention, failing loudly when it cannot fix.
|
|
12
|
+
|
|
13
|
+
sprint-doctor is that idea carried forward under a new name, with a stricter
|
|
14
|
+
decision/action split (zero-tool consult; the runner executes a bounded
|
|
15
|
+
action enum), an explicit environment-vs-engine-vs-task-shape triage, a
|
|
16
|
+
data-driven symptom/remedy registry, stagnation triggers on top of coach's
|
|
17
|
+
red-state events, and a generic engine-level pause/resume for human
|
|
18
|
+
hand-off. Every load-bearing coach idea was folded in and is credited
|
|
19
|
+
inline in the doctor doc: the three-layer error-handling hierarchy and
|
|
20
|
+
red-state taxonomy (its section 1), same-class-twice fingerprint cap,
|
|
21
|
+
no-recursion, interventions-as-review-evidence and deterministic test
|
|
22
|
+
stubbing (section 6), rescue-WIP-to-branch salvage (section 2.5), and the
|
|
23
|
+
consent-gated sanitized telemetry flow (section 4.4).
|
|
@@ -0,0 +1,256 @@
|
|
|
1
|
+
# Dolt manual recovery runbook (verified 2026-08-12, all steps live-tested on fleet-win-dev1)
|
|
2
|
+
|
|
3
|
+
This is the ground-truth procedure for standing up an ephemeral `dolt sql-server`
|
|
4
|
+
against a wedged embedded Dolt clone, querying/repairing it, and tearing it back
|
|
5
|
+
down -- verified end to end against a real clone on a real member, not written
|
|
6
|
+
from docs.
|
|
7
|
+
|
|
8
|
+
STATUS: this procedure is now AUTOMATED end to end by `settleDoltConflicts()`
|
|
9
|
+
(`packages/apra-fleet-se/fleet-sprint/dolt-settle.mjs`), which is the
|
|
10
|
+
executable form of this document and runs at both sync divergence terminals --
|
|
11
|
+
fleet members self-heal without a human or an LLM. The old Path A / Path B /
|
|
12
|
+
Tier 2 ladder and its LLM runbook (`dolt-tier2-runbook.md`) were retired and
|
|
13
|
+
deleted; see `dolt-sync-redesign.md`.
|
|
14
|
+
|
|
15
|
+
This document remains the SPECIFICATION that code encodes, and the human
|
|
16
|
+
copy-paste fallback for driving a clone by hand. Every step here was
|
|
17
|
+
independently rediscovered by trial and error at least once already (see
|
|
18
|
+
apra-fleet-ga61, apra-fleet-5mqg) -- which is exactly why it is written down
|
|
19
|
+
and encoded once.
|
|
20
|
+
|
|
21
|
+
## 0. Precondition: find the REAL data directory. Never hardcode it.
|
|
22
|
+
|
|
23
|
+
```
|
|
24
|
+
bd dolt status
|
|
25
|
+
```
|
|
26
|
+
|
|
27
|
+
Read the `Data:` line from the output (also mirrored in `.beads/metadata.json`'s
|
|
28
|
+
`dolt_mode` field). This is bd's own live, authoritative answer for THIS clone,
|
|
29
|
+
right now. Do not assume `.beads/embeddeddolt` -- that happens to be bd's
|
|
30
|
+
convention today, but a script that hardcodes it and finds a second/different
|
|
31
|
+
directory should treat that as orphaned residue from a prior failed recovery
|
|
32
|
+
(see apra-fleet-5mqg) and refuse to proceed blind, not silently pick one.
|
|
33
|
+
|
|
34
|
+
If `bd dolt status` reports `embedded (in-process, no server)`, there is no
|
|
35
|
+
live server to route through -- an ephemeral one must be started (step 2). If
|
|
36
|
+
it already reports a live server (host:port), skip straight to step 3 and
|
|
37
|
+
target that existing server/port instead of starting a second one.
|
|
38
|
+
|
|
39
|
+
## 1. Ensure the PINNED `dolt` binary is present (never a package manager)
|
|
40
|
+
|
|
41
|
+
The embedded `bd` binary does NOT ship enough of the Dolt CLI to drive a
|
|
42
|
+
merge-conflict resolution over SQL -- that requires the real `dolt` binary and
|
|
43
|
+
a real `dolt sql-server` process, temporarily, against the SAME data directory.
|
|
44
|
+
|
|
45
|
+
CORRECTED 2026-08-13 (`dolt-sync-redesign.md` Part 5, Precondition V2). This
|
|
46
|
+
step previously said `winget install --id DoltHub.Dolt` on Windows, plus
|
|
47
|
+
`apt`/`brew` on POSIX. That was a genuine mistake and must not be repeated:
|
|
48
|
+
winget delivered a STALE 1.86.3 build while this repo carries its own version
|
|
49
|
+
pin, so an OS package manager is a second, unpinned version channel -- banned
|
|
50
|
+
from every code path and runbook this design owns. It also created a PATH
|
|
51
|
+
problem of its own: a winget-installed binary is invisible to processes that
|
|
52
|
+
started before the install, which is every SSH-dispatched session.
|
|
53
|
+
|
|
54
|
+
The fleet's ONE pinned channel is the same GitHub release asset
|
|
55
|
+
`src/cli/dolt-install.ts` uses (`DOLT_VERSION`, currently `v2.2.0`), landed at
|
|
56
|
+
the fleet-managed path -- always referenced as an absolute path, never system
|
|
57
|
+
PATH, and needing no admin rights:
|
|
58
|
+
|
|
59
|
+
- Windows: `%USERPROFILE%\.apra-fleet\bin\dolt.exe`
|
|
60
|
+
- POSIX: `$HOME/.apra-fleet/bin/dolt`
|
|
61
|
+
|
|
62
|
+
Probe it first -- existence is not enough, it must actually RUN and report the
|
|
63
|
+
pinned version:
|
|
64
|
+
|
|
65
|
+
```bash
|
|
66
|
+
"$HOME/.apra-fleet/bin/dolt" version # POSIX
|
|
67
|
+
```
|
|
68
|
+
```powershell
|
|
69
|
+
& "$env:USERPROFILE\.apra-fleet\bin\dolt.exe" version # Windows
|
|
70
|
+
```
|
|
71
|
+
|
|
72
|
+
If it is missing, broken, or the wrong version, install it from the pinned
|
|
73
|
+
asset. `settleDoltConflicts()` does exactly this itself (`ensurePinnedDolt()`
|
|
74
|
+
in `dolt-settle.mjs`, design doc Part 5.3) -- members never run
|
|
75
|
+
`apra-fleet install`, so nothing else on a member ever puts a pinned dolt
|
|
76
|
+
there. By hand, the same sequence is:
|
|
77
|
+
|
|
78
|
+
**Windows:**
|
|
79
|
+
```powershell
|
|
80
|
+
New-Item -ItemType Directory -Force "$env:USERPROFILE\.apra-fleet\bin" | Out-Null
|
|
81
|
+
Invoke-WebRequest -Uri https://github.com/dolthub/dolt/releases/download/v2.2.0/dolt-windows-amd64.zip -OutFile "$env:TEMP\dolt-settle.zip" -TimeoutSec 300
|
|
82
|
+
Expand-Archive -Force "$env:TEMP\dolt-settle.zip" "$env:TEMP\dolt-settle"
|
|
83
|
+
Get-ChildItem -Recurse -Filter dolt.exe "$env:TEMP\dolt-settle" | Select-Object -First 1 |
|
|
84
|
+
Copy-Item -Force -Destination "$env:USERPROFILE\.apra-fleet\bin\dolt.exe"
|
|
85
|
+
```
|
|
86
|
+
|
|
87
|
+
**Linux / macOS** (swap the asset for `dolt-linux-amd64.tar.gz`,
|
|
88
|
+
`dolt-darwin-amd64.tar.gz` or `dolt-darwin-arm64.tar.gz` to match the member):
|
|
89
|
+
```bash
|
|
90
|
+
mkdir -p "$HOME/.apra-fleet/bin"
|
|
91
|
+
curl -fL --max-time 300 https://github.com/dolthub/dolt/releases/download/v2.2.0/dolt-linux-amd64.tar.gz -o /tmp/dolt-settle.tgz
|
|
92
|
+
mkdir -p /tmp/dolt-settle && tar -xzf /tmp/dolt-settle.tgz -C /tmp/dolt-settle
|
|
93
|
+
install -m 0755 "$(find /tmp/dolt-settle -type f -name dolt | head -1)" "$HOME/.apra-fleet/bin/dolt"
|
|
94
|
+
```
|
|
95
|
+
|
|
96
|
+
Then re-probe: the freshly installed binary must report the pinned version
|
|
97
|
+
before you continue. On Windows a replace can fail because a stray `dolt`
|
|
98
|
+
process holds the file -- kill any process whose executable path IS this
|
|
99
|
+
fleet-managed path (nothing else ever launches a binary there) and retry once.
|
|
100
|
+
|
|
101
|
+
Every `dolt ...` invocation below means that absolute pinned path, not a bare
|
|
102
|
+
`dolt` from PATH.
|
|
103
|
+
|
|
104
|
+
## 2. Start the ephemeral sql-server -- GENUINELY DETACHED
|
|
105
|
+
|
|
106
|
+
This is the step with real cross-platform divergence, and where most of the
|
|
107
|
+
danger lives (an interrupted/never-torn-down server is exactly what
|
|
108
|
+
apra-fleet-5mqg is about).
|
|
109
|
+
|
|
110
|
+
**Windows -- do NOT use `Start-Process` or `schtasks`:**
|
|
111
|
+
|
|
112
|
+
- `Start-Process -WindowStyle Hidden` looked detached but is NOT: it is still a
|
|
113
|
+
child of the invoking SSH/job-object session. Verified live: the sql-server
|
|
114
|
+
process (and its listening port) died the instant the launching
|
|
115
|
+
`execute_command` call's session ended, even with `-RedirectStandardOutput`/
|
|
116
|
+
`-RedirectStandardError` set and `-PassThru`.
|
|
117
|
+
- `schtasks`/`Register-ScheduledTask` + `Start-ScheduledTask` also failed:
|
|
118
|
+
verified live, `Get-ScheduledTaskInfo` reported `LastTaskResult 267011`
|
|
119
|
+
(`SCHED_S_TASK_QUEUED`) -- the task was queued but never actually ran,
|
|
120
|
+
because this is a non-interactive SSH session with no interactive logon
|
|
121
|
+
(the exact apra-fleet-i8qj failure mode, reproduced here independently).
|
|
122
|
+
- **What actually works, verified live:** spawn via WMI, which creates the
|
|
123
|
+
process under the WMI provider host's own session (session 0), fully
|
|
124
|
+
independent of the SSH session's job object:
|
|
125
|
+
|
|
126
|
+
```powershell
|
|
127
|
+
$result = Invoke-CimMethod -ClassName Win32_Process -MethodName Create -Arguments @{
|
|
128
|
+
CommandLine = '"C:\Users\<you>\.apra-fleet\bin\dolt.exe" sql-server --host 127.0.0.1 --port 13399 --data-dir .beads/embeddeddolt'
|
|
129
|
+
CurrentDirectory = 'C:\akhil\git\apra-fleet'
|
|
130
|
+
}
|
|
131
|
+
$result.ProcessId # <-- record this PID for teardown (step 5)
|
|
132
|
+
```
|
|
133
|
+
Verify it is actually up before proceeding: `Get-Process -Id <pid>` should
|
|
134
|
+
show it running under `SI` (session ID) `0`, and
|
|
135
|
+
`Test-NetConnection -ComputerName 127.0.0.1 -Port 13399` should report
|
|
136
|
+
`TcpTestSucceeded: True`.
|
|
137
|
+
|
|
138
|
+
**Linux / macOS (POSIX):** the SSH-session-job-object problem does not exist on
|
|
139
|
+
POSIX in the same way; the existing `nohup ... & disown` pattern (as used by
|
|
140
|
+
`task-wrapper.ts`'s bash wrapper for `long_running` commands elsewhere in this
|
|
141
|
+
codebase) is sufficient:
|
|
142
|
+
|
|
143
|
+
```bash
|
|
144
|
+
nohup dolt sql-server --host 127.0.0.1 --port 13399 --data-dir .beads/embeddeddolt \
|
|
145
|
+
> /tmp/dolt-recovery.log 2>&1 &
|
|
146
|
+
echo "PID:$!"
|
|
147
|
+
disown
|
|
148
|
+
```
|
|
149
|
+
|
|
150
|
+
## 3. Connect and query -- the exact working flag set (verified live)
|
|
151
|
+
|
|
152
|
+
```
|
|
153
|
+
dolt --no-tls --host=127.0.0.1 --port=13399 sql -q "USE beads; SELECT ...;"
|
|
154
|
+
```
|
|
155
|
+
|
|
156
|
+
Two specific, non-obvious landmines, both reproduced live on this run:
|
|
157
|
+
|
|
158
|
+
1. **`--no-tls` must come BEFORE `--host`/`--port`.** With `--host`/`--port`
|
|
159
|
+
given first and `--no-tls` after, the client still requested TLS and failed
|
|
160
|
+
with `TLS requested but server does not support TLS` -- Dolt's global-flag
|
|
161
|
+
parser is order-sensitive here. Always lead with `--no-tls`.
|
|
162
|
+
2. **Do NOT pass `--user=root` or `--password` at all**, even though root with
|
|
163
|
+
an empty password is the documented default. Explicitly passing `--user=root`
|
|
164
|
+
switches the client into a credential-resolution path that issues an
|
|
165
|
+
interactive `Enter password:` prompt and then fails non-interactively with
|
|
166
|
+
`Failed to parse credentials: The handle is invalid` -- this is the EXACT
|
|
167
|
+
apra-fleet-ga61 symptom, reproduced live and now root-caused precisely: it
|
|
168
|
+
is triggered by explicitly specifying `--user`/`--password`, not by network
|
|
169
|
+
mode itself. Omitting both flags entirely lets the client silently
|
|
170
|
+
authenticate as root with the empty-password default with zero prompt.
|
|
171
|
+
(Separately, on Windows/PowerShell specifically: an empty-string argument
|
|
172
|
+
like `--password ''` can be silently dropped by PowerShell's native-argument
|
|
173
|
+
passing, shifting the NEXT flag into becoming its value -- another reason to
|
|
174
|
+
just omit the flag rather than try to pass an explicit empty value.)
|
|
175
|
+
|
|
176
|
+
Use `USE beads;` as the first statement of every query -- the target database
|
|
177
|
+
name is not implied by `--data-dir` alone once connected over SQL.
|
|
178
|
+
|
|
179
|
+
The actual recovery SQL (identical across all three OSes -- this part has no
|
|
180
|
+
platform variance):
|
|
181
|
+
|
|
182
|
+
```sql
|
|
183
|
+
SET @@dolt_allow_commit_conflicts = 1;
|
|
184
|
+
CALL DOLT_MERGE('origin/main');
|
|
185
|
+
SELECT `table` FROM dolt_conflicts; -- enumerate EVERY conflicted table
|
|
186
|
+
CALL DOLT_CONFLICTS_RESOLVE('--theirs', 'issues'); -- table name is REQUIRED
|
|
187
|
+
CALL DOLT_COMMIT('-m', 'manual recovery: resolve conflict');
|
|
188
|
+
SELECT COUNT(*) FROM dolt_conflicts; -- MUST be 0 before you republish
|
|
189
|
+
```
|
|
190
|
+
A bare `--theirs` is only correct when discarding our side's version of those
|
|
191
|
+
rows is genuinely acceptable. `settleDoltConflicts()` applies the full rulebook
|
|
192
|
+
instead -- per-field last-writer-wins by `updated_at` for `issues` and
|
|
193
|
+
`dependencies`, a real set-union INSERT for `labels`, generic
|
|
194
|
+
LWW-or-theirs for anything else (design doc Part 3.2 step 4). Resolving by hand
|
|
195
|
+
with real content on both sides, follow the same rules rather than clobbering
|
|
196
|
+
one side.
|
|
197
|
+
|
|
198
|
+
Then, outside the SQL session -- and ONLY AFTER the teardown in step 5 --
|
|
199
|
+
`bd dolt pull` (verify), `bd dolt push` (republish). The order matters; see
|
|
200
|
+
step 5.
|
|
201
|
+
|
|
202
|
+
## 4. Verify before teardown
|
|
203
|
+
|
|
204
|
+
```
|
|
205
|
+
dolt --no-tls --host=127.0.0.1 --port=13399 sql -q "USE beads; SELECT * FROM dolt_status;"
|
|
206
|
+
```
|
|
207
|
+
Confirm no residual uncommitted/conflicted state before tearing the server down.
|
|
208
|
+
|
|
209
|
+
## 5. Teardown -- ALWAYS, every path, even on failure, and BEFORE republishing
|
|
210
|
+
|
|
211
|
+
Ordering, corrected (design doc Part 7.2): tear the ephemeral server down
|
|
212
|
+
FIRST, then run `bd dolt pull` / `bd dolt push`. Embedded-mode `bd` must open
|
|
213
|
+
the same data dir the server still holds under Dolt's per-directory exclusive
|
|
214
|
+
lock, so republishing while the server is still up fails or falls back
|
|
215
|
+
read-only.
|
|
216
|
+
|
|
217
|
+
**Windows:**
|
|
218
|
+
```powershell
|
|
219
|
+
Stop-Process -Id <pid> -Force -ErrorAction SilentlyContinue
|
|
220
|
+
```
|
|
221
|
+
**POSIX:**
|
|
222
|
+
```bash
|
|
223
|
+
kill <pid>
|
|
224
|
+
```
|
|
225
|
+
|
|
226
|
+
Then, on every OS:
|
|
227
|
+
- Verify the port is actually closed (`Test-NetConnection` / `nc -z` /
|
|
228
|
+
`lsof -i:<port>`).
|
|
229
|
+
- Verify `bd dolt status` reports `embedded (in-process, no server)` again and
|
|
230
|
+
`.beads/metadata.json`'s `dolt_mode` is `"embedded"` -- if this recovery
|
|
231
|
+
procedure ever explicitly flips `dolt_mode` to `"server"` as part of routing
|
|
232
|
+
bd itself through the ephemeral server (this manual run did NOT do that --
|
|
233
|
+
it talked to the raw `dolt` CLI directly, bypassing bd entirely -- and
|
|
234
|
+
`settleDoltConflicts()` deliberately does the same, which is what makes its
|
|
235
|
+
teardown unconditionally safe), flipping it back is
|
|
236
|
+
MANDATORY and must happen in a `finally`/guaranteed-cleanup block, not as
|
|
237
|
+
the last step of a list that can be abandoned mid-way (exactly what went
|
|
238
|
+
wrong in the incident apra-fleet-5mqg documents).
|
|
239
|
+
- Remove any temp log files created for the ephemeral server's stdout/stderr.
|
|
240
|
+
|
|
241
|
+
## Why this matters for the code fix (apra-fleet-5mqg, apra-fleet-ga61)
|
|
242
|
+
|
|
243
|
+
Every numbered landmine above (`--no-tls` ordering, omitting `--user`/
|
|
244
|
+
`--password`, WMI vs `Start-Process`/`schtasks` for real detachment on Windows,
|
|
245
|
+
never hardcoding the data directory) was independently rediscovered by trial
|
|
246
|
+
and error here, and was ALSO independently rediscovered by trial and error by
|
|
247
|
+
the Tier-2 LLM dispatch whose transcript motivated apra-fleet-5mqg. That was
|
|
248
|
+
strong evidence this entire procedure is mechanical and fully scriptable --
|
|
249
|
+
there was no judgment call anywhere in this runbook, only flag/ordering
|
|
250
|
+
trivia a fixed script encodes once and never has to rediscover again. That
|
|
251
|
+
script now exists: `fleet-sprint/dolt-settle.mjs`.
|
|
252
|
+
|
|
253
|
+
CAVEAT on the evidence: every landmine above was verified live against a
|
|
254
|
+
winget-installed dolt **1.86.3**, not against the pinned **v2.2.0** the fleet
|
|
255
|
+
now installs. Re-verifying them on the pinned binary, per OS, is Precondition
|
|
256
|
+
V0 of the design doc's Part 4 verification pass.
|