openclaw-code-agent 4.7.12 → 4.7.14

package/README.md

@@ -128,15 +128,15 @@ Because worktree isolation defaults to `delegate`, `defaultWorkdir` should norma
 
 Chat-launched sessions route updates back to their originating chat thread. For agent-launched tool sessions without an origin route, configure `fallbackChannel` or `agentChannels` in the reference guide.
 
-The current package requires, is built against, and is validated against OpenClaw `2026.9.1`, while its plugin API, Gateway, and peer dependency contracts retain the verified `2026.8.1` compatibility floor. The manifest still declares `contracts.tools`, the plugin still imports only `openclaw/plugin-sdk/plugin-entry`, and its session store, wake routing, callbacks, worktree flows, and Codex/Claude model restrictions remain plugin-owned. OpenClaw's host-side model catalog and activatable-harness checks do not rewrite Code Agent's unprefixed Codex harness models (`gpt-5.6-sol`, `gpt-5.6-terra`, and `gpt-5.6-luna`), its Claude Code aliases (`sonnet` and `opus`), or either harness-scoped allowlist. Existing sessions, permissions, Telegram/topic routes, automation delivery context, callback ownership, and namespaced tool allowlists remain under the same plugin contracts. OpenClaw `2026.9.1` improves origin-channel approvals, callback acknowledgement, cron delivery recovery, managed worktrees, and Codex lifecycle behavior; these are host-side improvements and do not transfer ownership of Code Agent's routing or follow-through state. Installing this package performs no host-side configuration migration.
+The current package requires, is built against, and is validated against OpenClaw `2026.9.2`, while its plugin API, Gateway, and peer dependency contracts retain the verified `2026.8.1` compatibility floor. The manifest still declares `contracts.tools`, the plugin still imports only `openclaw/plugin-sdk/plugin-entry`, and its session store, wake routing, callbacks, worktree flows, and Codex/Claude model restrictions remain plugin-owned. OpenClaw's host-side model catalog and activatable-harness checks do not rewrite Code Agent's unprefixed Codex harness models (`gpt-6-astra`, `gpt-5.6-sol`, `gpt-5.6-terra`, and `gpt-5.6-luna`), its Claude Code aliases (`sonnet` and `opus`), or either harness-scoped allowlist. Existing sessions, permissions, Telegram/topic routes, automation delivery context, callback ownership, and namespaced tool allowlists remain under the same plugin contracts. OpenClaw `2026.9.2` improves delegated approval outcomes, automation topic routing, and reply recovery; these are host-side improvements and do not transfer ownership of Code Agent's routing or follow-through state. Installing this package performs no host-side configuration migration. OpenClaw 2026.9.2 broadens default cross-agent session visibility; see [SDK readiness and migration guidance](docs/REFERENCE.md#openclaw-202692-sdk-readiness) for optional isolation settings and harness configuration examples.
 
-If you use Codex, make sure the local `codex` command or `OPENCLAW_CODEX_APP_SERVER_COMMAND` override is available and authenticated. Codex-specific defaults live under `harnesses.codex`: `reasoningEffort` is sent as `reasoningEffort`, and `fastMode: true` sends `service_tier: "fast"` on Codex App Server thread, resume, and turn payloads. The Codex harness starts the app server with stdio listener args by default, only sends UUID-shaped backend thread IDs to `thread/resume`, and reports startup timeouts with redacted recent stderr. When Codex auth is inconsistent, this is the recommended `~/.codex/config.toml` setting:
+If you use Codex, make sure the local `codex` command or `OPENCLAW_CODEX_APP_SERVER_COMMAND` override is available and authenticated. GPT-6 Astra is the default Codex model; GPT-5.6 Sol, Terra, and Luna remain supported explicit overrides. Codex-specific defaults live under `harnesses.codex`: `reasoningEffort` is sent as `reasoningEffort`, and `fastMode: true` sends `service_tier: "fast"` on Codex App Server thread, resume, and turn payloads. The Codex harness starts the app server with stdio listener args by default, only sends UUID-shaped backend thread IDs to `thread/resume`, and reports startup timeouts with redacted recent stderr. When Codex auth is inconsistent, this is the recommended `~/.codex/config.toml` setting:
 
 ```toml
 forced_login_method = "chatgpt"
 ```
 
-For Codex sessions authenticated with an OpenAI API key, the plugin estimates token charges from the App Server's per-response usage and shows the cumulative amount in session listings and `agent_stats`. ChatGPT OAuth/subscription sessions remain `$0` because they are quota-backed rather than pay-as-you-go API calls. The estimator covers the built-in GPT-5.6 Sol, Terra, and Luna models, including cached input, cache writes, Fast mode, and long-context rates; unknown models remain unpriced. Reasoning tokens are already included in App Server `outputTokens` and are not added a second time.
+For Codex sessions authenticated with an OpenAI API key, the plugin estimates token charges from the App Server's per-response usage and shows the cumulative amount in session listings and `agent_stats`. ChatGPT OAuth/subscription sessions remain `$0` because they are quota-backed rather than pay-as-you-go API calls. The estimator covers the built-in GPT-6 Astra and GPT-5.6 Sol, Terra, and Luna models, including cached input, cache writes, Fast mode, and long-context rates; unknown models remain unpriced. Reasoning tokens are already included in App Server `outputTokens` and are not added a second time.
 
 If you use experimental OpenCode, make sure local `opencode >= 1.16.2` is available and configured with provider auth. The plugin starts `opencode serve` per session on localhost and uses OpenCode's classic session lifecycle routes while v2 session wait remains unavailable. Leave `harnesses.opencode.defaultModel` unset to let OpenCode choose its configured provider default, or pass an explicit `provider/model` string for a launch.
 
@@ -195,6 +195,8 @@ Merge, PR, ordinary terminal, and no-change worktree outcomes use a two-step com
 ### Goal Tasks
 
 Goal tasks are explicit autonomous loops for work that should keep iterating toward a defined finish line. They do not replace ordinary coding sessions.
+Session notification headings include a concise `| reasoning: medium` field when OCA knows the selected harness/model supports the session's reasoning setting. `agent_launch(reasoning_effort="high")` overrides the harness default for that launch; resume and fork retain the saved setting unless overridden. Launch, approval/progress, terminal and later worktree notifications use that session setting even after a Gateway restart or default change. Unknown historical settings and unsupported models/harnesses (including OpenCode, which does not forward this option) omit the field. OCA does not infer a backend default or a silently downgraded Claude effort level.
+
 Goal iteration progress is controller progress: the counter advances only when the goal controller starts another agent turn after a missing completion promise or failed verifier. If the agent performs several review/implementation passes inside one successful turn, those internal passes should appear in the single completion summary rather than as separate goal iteration notifications.
 
 Ask in normal chat:

package/dist/chunks/agent-pr.js

@@ -1 +1 @@
-import{Ba as a,Ca as b,Da as c,Ea as d,Fa as e,Ga as f,Ha as g,Ia as h,Ja as i,Ka as j,La as k,Ma as l,Na as m,Oa as n,Pa as o}from"./chunk.js";export{m as buildPrCompletionWakeOutcomeKey,a as buildPrMetadata,l as buildPrOutcomeDetailLines,b as createRuntimePrMetadataProvider,j as discoverExistingTargetPr,f as formatPrBody,d as isOcaFallbackPrBody,c as isOcaGeneratedPrBody,e as isOcaGeneratedPrTitle,o as makeAgentPrTool,h as normalizeForceNewReplacementPrStatus,n as refreshOpenPrMetadata,i as resolveExistingTargetPrUpdateBranch,k as resolveExistingTargetPrUpdateSourceBranch,g as shouldIgnoreClosedTargetPrForForceNew};
+import{Da as a,Ea as b,Fa as c,Ga as d,Ha as e,Ia as f,Ja as g,Ka as h,La as i,Ma as j,Na as k,Oa as l,Pa as m,Qa as n,Ra as o}from"./chunk.js";export{m as buildPrCompletionWakeOutcomeKey,a as buildPrMetadata,l as buildPrOutcomeDetailLines,b as createRuntimePrMetadataProvider,j as discoverExistingTargetPr,f as formatPrBody,d as isOcaFallbackPrBody,c as isOcaGeneratedPrBody,e as isOcaGeneratedPrTitle,o as makeAgentPrTool,h as normalizeForceNewReplacementPrStatus,n as refreshOpenPrMetadata,i as resolveExistingTargetPrUpdateBranch,k as resolveExistingTargetPrUpdateSourceBranch,g as shouldIgnoreClosedTargetPrForForceNew};