chore(release): reflexio-ai v0.2.14#29
Closed
yilu331 wants to merge 13 commits into
Closed
Conversation
…install - Add --force-extraction to session-end publish so extraction runs regardless of batch-interval throttling - Redirect publish output to ~/.reflexio/logs/stop-hook.log instead of silently discarding errors - Remove dead stop_hook_active guard (field not in Claude Code payload) - Auto-install Stop hook when expert mode is enabled via setup wizard - Update HOOK.md to reflect all changes
Add 5 high-signal patterns that trigger auto-publish without user input: - Self-recovered tool failures (error + recovery arc) - Retry chains (2+ failures on same operation) - Documentation/behavior gaps (docs say X, reality is Y) - Workarounds for limitations (used alternative approach) - Anomalous results (unexpected zeros, row-count mismatches) Also add end-of-session extraction nudge: suggest /reflexio-extract when session had any friction points.
- test_normal_mode_no_stop_hook: verifies Stop hook absent without --expert - test_expert_mode_installs_stop_hook: verifies Stop hook present with --expert, checks command points to handler.js - test_expert_mode_stop_hook_idempotent: verifies no duplication on repeated install - test_remove_from_dir_cleans_stop_hook: verifies uninstall removes the Stop hook
Stop fires once per turn (every assistant response), causing repeated publishes of the growing transcript and wasteful --force-extraction on every turn. SessionEnd fires once when the session exits — correct for full-session capture. - _merge_hook_config: Stop → SessionEnd - _remove_hook_config: add SessionEnd to removal list (keep Stop for backwards compat with old installs) - handler.js: update JSDoc comments - HOOK.md: update event name, installation example, frontmatter - tests: rename and update all SessionEnd assertions
Root cause: description led with "Search" which framed Reflexio as a search tool, causing Claude to route corrections to its built-in memory system instead of triggering the Reflexio publish flow. Changes: - Lead description with "Publish user corrections" (the missed trigger) - Add when_to_use field with concrete trigger phrases - Explain profile vs playbook: facts publish immediately, procedures publish after user confirmation (explicit or implicit) - Distinguish from Claude's built-in local memory - Add callout in skill body preventing memory/Reflexio conflation - Reorder trigger list: corrections first, search second
Root cause: skill tried to handle both when-to-trigger and how-to-publish, but Claude decides whether to invoke a skill before reading its content. Triggering guidance needs to be in rules (always loaded). Changes: - rules/reflexio.md: add "When to Publish to Reflexio" section with profile/playbook/procedure guidance and confirmation model; soften existing section titles from commanding to guiding tone - skill/SKILL-expert.md: simplify from 235 to 98 lines; keep only how-to (payload formats, correction patterns, end-of-session nudge) - skill/SKILL.md: update description to triggering conditions only per skill creation guidelines - skill/references/: new directory with proactive-patterns.md and server-management.md (heavy content moved out of main skill) - setup_cmd.py: copy references/ dir in expert mode - tests: add test_expert_mode_installs_references and test_normal_mode_no_references
delete_all_*, delete_*_by_ids, delete_*, update_*, and publish_*
methods now clear relevant cache entries so subsequent list/get
calls return fresh data instead of stale cached results.
Changes:
- cache.py: add clear() and invalidate(method_name) methods;
store method_name in each entry for prefix-based invalidation
- client.py: call cache.invalidate("get_profiles") on profile
mutations, cache.invalidate("get_agent_playbooks") on playbook
mutations, cache.clear() on nuclear deletes (delete_all_interactions,
delete_all_playbooks)
- tests: 7 new tests for clear, invalidate, and client integration
reflexio interactions list without --user-id failed with a Pydantic validation error because GetInteractionsRequest requires user_id. Now uses require_user_id() to fall back to REFLEXIO_USER_ID env var, matching the pattern used by other list commands.
- setup_cmd.py: shell-quote all hook paths with shlex.quote for paths with spaces; remove SessionEnd hook + expert artifacts (references/, reflexio-extract command) when re-installing in normal mode; extract _remove_reflexio_hook helper - handler.js: use positional args ($1..$5) instead of template literal interpolation in sh -c to prevent shell injection - HOOK.md: fix inaccurate docs — tool_use blocks ARE preserved as tools_used metadata, not skipped - server-management.md: add language identifier to code fence - tests: add test_normal_reinstall_removes_expert_artifacts
|
Caution Review failedPull request was closed or merged during review 📝 WalkthroughWalkthroughThis release adds version 0.2.14 with cache invalidation management, expert mode setup support for Claude Code hooks, SessionEnd event integration, and comprehensive documentation updates. Changes include cache clearing/invalidation methods, client-side cache management after mutations, enhanced CLI setup logic for expert configuration, and updated hook and skill documentation. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
Collaborator
Author
|
Version bump pushed directly to main in 9813a8a. PR no longer needed. |
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.
Summary
Version bump: reflexio-ai 0.2.13 → 0.2.14
What's in 0.2.14
SessionEnd(once per session) instead ofStop(per-turn);--force-extraction; logging to~/.reflexio/logs/stop-hook.log; shell-safe positional argsrules/reflexio.md(always loaded); skill simplified from 235 → 98 lines; references extracted toreferences/clear()andinvalidate()methods; 12 mutation methods now invalidate stale cachereflexio interactions listfalls back toREFLEXIO_USER_IDenv varTest plan
reflexio --version→ 0.2.14pip install reflexio-ai==0.2.14→ OKSummary by CodeRabbit
New Features
Bug Fixes
Documentation