Conversation
Frames injected memory with an identity-binding header (naming the peer when resolvable) and closes it with an applying-memory directive, so a third-person representation is not read as notes about someone else and stated preferences are not dropped as out of scope. Adds the injection.explicitObservations mode (strip | keep) controlling whether the raw timestamped explicit-observations wall reaches the agent. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Important Review skippedAuto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
…eugene/first_turn_prompt_configuration
૮ >ﻌ< ა ci reviewran on a7b0363 — Merge remote-tracking branch 'origin/eugene/fix-honcho_loggi ❌ Job failuresCheck contributors / check-attribution · View jobJob Check contributors / check-attribution failed. debug infoCI timingsCI timings · View report · View jobWall time 1445m1s vs 1440m46s (+0.3%). 8 job(s) slower, 3 faster, 1 unchanged.
|
ajspig
pushed a commit
that referenced
this pull request
Sep 18, 2026
`_TERMINAL_KANBAN_TOOLS` listed only `kanban_complete` / `kanban_block`, so the turn-end guard fired at workers that had already handed the card off correctly: - A build worker that calls `kanban_request_review` moves the card from `running` to `review` (tools/kanban_tools.py `_handle_request_review`), yet `session_called_kanban_terminal()` returned False and the nudge told it "Task is still `running`" — false by then — and to call `kanban_complete`, which would close a card that must go through review. Both goal-mode prompts name that tool explicitly (hermes_cli/goals.py `KANBAN_GOAL_CONTINUATION_TEMPLATE` and `KANBAN_GOAL_FINALIZE_TEMPLATE`), so the prompt and the guard contradicted each other. - Review agents hit the same wall. The review lane spawns through the same `_default_spawn`, which sets `HERMES_KANBAN_TASK`, so the guard is active for them, and the force-loaded sdlc-review skill's decision table ends the request-changes path on `kanban_request_changes`. Adds both handoff tools to the set. Workers that obey their prompt now exit cleanly; the guard still fires for a non-terminal board tool, which is the case it exists for. Covers lifecycle mismatch #2 of NousResearch#94916 only — the `dispatch --dry-run` half is a separate change and is discussed on the issue.
ajspig
pushed a commit
that referenced
this pull request
Sep 18, 2026
… in the flag table - Dialog 2 now shows the read-then-answer step for one benign prompt and warns against blind timed Enter, and names --permission-mode acceptEdits as the narrower opt-in (idea from PR NousResearch#113456). - Quick Reference row and pitfall #2 carry the opt-in wording instead of teaching Down+Enter as the expected move. - Regenerated the claude-code docs page.
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.
What does this PR do?
Shapes the first-turn memory prompt so injected Honcho memory is read as standing requirements from the current user rather than as background about a third party.
Stacked on
eugene/fix-honcho_logging(config-key repairs); this PR uses the session-start rendering added there. Three changes:## Who this memory is aboutheader naming the peer when it can be resolved, so a third-person representation ("mei wants …") is not misread as notes about somebody else.## Applying this memory, stating that the preferences above are current requirements, not scoped to the project they were stated in, to be reproduced literally where they name a marker or format, and that they outrank the agent's own style defaults. Appended only when the payload actually carries preferences (a card-only or summary-only injection states none), and after budget truncation so it can't be the part that gets cut.injection.explicitObservations(stripdefault |keep).stripdrops the raw timestamped explicit-observations wall from the peer representation and injects only the distilled deductions/patterns;keepinjects the representation as rendered. The choice is load-bearing: a requirement the user stated outright lives in the explicit layer, and the deriver does not always roll it up into a deduction.Related Issue
No open issue.
Fixes #
Type of Change
Changes Made
plugins/memory/honcho/__init__.py_MEMORY_SUBJECT_HEADER[_NAMED]and_subject_header(): identity-binding preamble prepended to the session-start context; falls back to an unnamed binding rather than naming the wrong peer._MEMORY_DIRECTIVEappended inprefetchwhen the injection carries a user representation or dialectic result.injection.explicitObservationsresolver,_strip_explicit_observations(), and the strip branch in_format_first_turn_context. Unknown values fall back to the default with a warning rather than raising; non-string values pass through untouched.plugins/memory/honcho/session.pysubject_peer_id(): resolves the peer a session's memory is about without forcing session creation.tests/honcho_plugin/test_session.pyTestExplicitObservationsModecovering the default, both modes, unknown-value fallback, case/whitespace normalisation, and the empty-after-strip edge case.How to Test
pytest tests/honcho_plugin/ -q— all tests pass, includingTestExplicitObservationsMode.HONCHO_LOGGING=1(from the base PR), inspect a turn-1 record in~/.honcho/injection.log: the payload opens with## Who this memory is aboutand, when a user representation is present, closes with## Applying this memory.injection.explicitObservations: keepand confirm the## Explicit Observationssection reappears in the payload; withstrip(default) it is absent while## Deductive Observationsremains.Checklist
Code
fix(scope):,feat(scope):, etc.)pytest tests/ -qand all tests passDocumentation & Housekeeping
docs/, docstrings) — or N/A (docstrings; behavior config lives in~/.honcho/config.json)cli-config.yaml.exampleif I added/changed config keys — or N/A (new key is Honcho plugin config in~/.honcho/config.json)CONTRIBUTING.mdorAGENTS.mdif I changed architecture or workflows — or N/AScreenshots / Logs
Shape of a turn-1 injection with a user representation present: