✨ feat(agent-signal): add execAgent plumbing for self-iteration migration#15187
Merged
Conversation
…tion Phase 1 of LOBE-9434: introduces dormant plumbing for converging agent execution onto execAgent. No behavior changes for any existing caller — every piece is a no-op until later phases wire it up. - Add `ExecAgentAppContext.suppressSignal` flag and `sourceMessageId` - Add `shouldSuppressSignal` helper; gate the `agent.user.message` re-emission in `aiAgent.execAgent` so future builtin/background runs cannot recurse into the analyzeIntent pipeline - Register `self-iteration` builtin agent + `SELF_ITERATION_AGENT_SLUGS` - Add `finalStateExtractor` (`extractFromFinalState` / `extractMutations` / `extractArtifacts`) for reading tool-result kind partitions off a persisted AgentState snapshot - Register a no-op `completionPolicy` listener on `agent.execution.completed` with an optional `onSelfIterationCompleted` callback (undefined by default) Tests: 17 new unit tests across suppressSignal, finalStateExtractor, and completionPolicy.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 5d483c4f9b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
|
||
| results.push({ | ||
| apiName: typeof message.apiName === 'string' ? message.apiName : undefined, | ||
| data: contentRecord ?? content, |
There was a problem hiding this comment.
Preserve pluginState payload for extracted tool results
When kind is discovered via pluginState.kind (because message.content is plain text/non-JSON), extractFromFinalState still sets data to contentRecord ?? content, which drops the structured pluginState fields (e.g., IDs needed for mutation/artifact projection). In that scenario downstream self-iteration completion handlers will receive only a string message instead of the actual tool result object, making final-state reconstruction incomplete.
Useful? React with 👍 / 👎.
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## canary #15187 +/- ##
=========================================
Coverage 70.92% 70.93%
=========================================
Files 3149 3152 +3
Lines 313896 313971 +75
Branches 33311 28558 -4753
=========================================
+ Hits 222628 222705 +77
+ Misses 91100 91098 -2
Partials 168 168
Flags with carried forward coverage won't be shown. Click here to find out more.
🚀 New features to boost your workflow:
|
This was referenced May 25, 2026
hardy-one
pushed a commit
to hardy-one/lobehub
that referenced
this pull request
May 27, 2026
…tion (lobehub#15187) Phase 1 of LOBE-9434: introduces dormant plumbing for converging agent execution onto execAgent. No behavior changes for any existing caller — every piece is a no-op until later phases wire it up. - Add `ExecAgentAppContext.suppressSignal` flag and `sourceMessageId` - Add `shouldSuppressSignal` helper; gate the `agent.user.message` re-emission in `aiAgent.execAgent` so future builtin/background runs cannot recurse into the analyzeIntent pipeline - Register `self-iteration` builtin agent + `SELF_ITERATION_AGENT_SLUGS` - Add `finalStateExtractor` (`extractFromFinalState` / `extractMutations` / `extractArtifacts`) for reading tool-result kind partitions off a persisted AgentState snapshot - Register a no-op `completionPolicy` listener on `agent.execution.completed` with an optional `onSelfIterationCompleted` callback (undefined by default) Tests: 17 new unit tests across suppressSignal, finalStateExtractor, and completionPolicy.
arvinxx
added a commit
that referenced
this pull request
Jun 1, 2026
…nt slugs (#15202) The Phase 1 consolidation into a single `self-iteration` slug (PR #15187, inheriting commit 627f899 from the closed #15116) conflated three distinct background flows that have: - Independent receipt tables and idempotency Redis namespaces - Different preflight / brief projection paths - Different audit pipelines `one identifier = one behavior` is a load-bearing contract once these agents are routed through the standard execAgent plugin lookup. Restore the 3 mode-specific slugs so each agent declares its own tool surface: | slug | future plugin identifier | | ----------------------- | ------------------------------- | | `nightly-review` | `agent-signal-review` | | `self-reflection` | `agent-signal-reflection` | | `self-feedback-intent` | `agent-signal-feedback-intent` | `SELF_ITERATION_AGENT_SLUGS` now contains all three; `completionPolicy` dispatches on slug membership rather than equality; callback receives the resolved `agentId` so mode-specific bookkeeping can route from it. Plugin arrays reference the future identifiers but the tool packages are not yet registered — invoking any of these agents today runs the LLM with no tools (dormant by design). Tool-package registration follows in a separate PR. No behavior change for existing callers (none invoke these slugs yet).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
💻 Change Type
🔗 Related Issue
Part of LOBE-9434 (parent).
Relates to #15116 — this is Phase 1 (dormant plumbing) of the staged split for that PR.
🔀 Description of Change
Phase 1 of LOBE-9434: introduce dormant infrastructure for converging
agent execution onto
execAgent. Every piece in this PR is a no-op for everyexisting caller — later phases will start using these primitives.
The goal of splitting Phase 1 out is to let the high-risk pieces of #15116
(switching
userMemoryto asyncexecAgent, deletingexecuteSync,migrating
executeSelfIteration) land in much smaller, individuallyrevertable PRs once this foundation has baked.
Contents:
ExecAgentAppContext.suppressSignalflag +sourceMessageId;shouldSuppressSignalhelper; gate theagent.user.messagere-emission insideAiAgentService.execAgentsuppressSignal: trueand no caller passes theself-iterationslug yet — current behaviour identical.self-iterationbuiltin agent +SELF_ITERATION_AGENT_SLUGSsetshouldSuppressSignal.finalStateExtractor(extractFromFinalState/extractMutations/extractArtifacts)completionPolicyregistered onagent.execution.completed,onSelfIterationCompletedcallback isundefinedby defaultDifferences vs. the original #15116:
shouldSuppressSignalis actually wired intoAiAgentService.execAgent(the original PR added the helper but never called it).packages/builtin-agents/src/types.tsorpackages/builtin-agents/src/index.ts.log(…)debug logger inaiAgent/index.tsfor the enqueue catch handler (original PR switched toconsole.error).completionPolicyshort-circuits before thetry { … }block when no callback is provided, instead of always entering it.🧪 How to Test
bun run type-checkclean for everything Phase 1 touches (remaining errors are pre-existing cloud-business issues unrelated to this PR)src/server/services/agentSignal/__tests__/suppressSignal.test.ts(5 tests)src/server/services/agentSignal/services/selfIteration/__test__/finalStateExtractor.test.ts(6 tests)src/server/services/agentSignal/policies/completionPolicy.test.ts(6 tests)packages/builtin-agents,policies/, andservices/selfIteration/: 300 tests pass.📝 Additional Information
No behaviour change. Safe to merge first; subsequent phases:
executeViaExecAgent(still not wired into any caller)userMemory.ts(the high-risk piece), then migrateexecuteSelfIterationcallers one at a timeonSelfIterationCompletedcallback and clean up legacyexecute.tsCloud impact: none —
aiAgent/index.tsis an open-source path; the suppress check is gated byappContext?.suppressSignal === trueor membership inSELF_ITERATION_AGENT_SLUGS, neither of which is reachable by any current cloud caller.