🦛 Hippo Memory Insights — 2026-05-18

[github-actions[bot]]

Summary

• Memories in store: 531 (531/531 embedded, diverse topics, mostly fresh <7d, high avg strength)
• New lessons learned from commits today: 0 (no fix/revert/bug commits in the last 7 days)
• Consolidation run: 79 episodic → semantic merges, 29 new semantic patterns, 29 duplicates deduped, 174 auto-shared to global store
• 7 low-quality memories flagged by audit (run hippo audit to review)
• Highest-confidence memory: mem_56999cd476c3 (score 0.988) — stale .lock.yml CI incident

Top Memories Surfaced

1. mem_56999cd476c3 ✅ score=0.988 — Recurring incident: stale workflow .lock.yml files cause churn and CI friction; run make recompile after any markdown workflow edit.
2. mem_de1a9a9c2a51 score=0.658 — CI WILL FAIL if pre-commit formatting/lint check is skipped — this is automatic and non-negotiable.
3. mem_7cfbbf7ec61e score=0.685 — The validate-yaml job in .github/workflows/ci.yml scans all YAML files for ANSI escape sequences before other jobs run.
4. mem_d9d310064f1e score=0.801 — skills/javascript-refactoring/SKILL.md — Guide for refactoring JavaScript into separate .cjs files.
5. mem_67941271277f score=0.757 — scratchpad/validation-refactoring.md — Step-by-step refactoring guide for validation complexity.
6. mem_0d2a2d2474a0 score=0.582 — Pre-commit formatting failures have caused 5 CI failures in a single day — make fmt before every commit.
7. mem_79b981ac6a40 score=1.00 — Architecture note: gh-aw compiles markdown workflows into GitHub Actions; it is not the GitHub Copilot CLI.

Suggested Improvements

Error Patterns

• Stale .lock.yml files (highest-confidence verified incident): Always run make recompile after editing any markdown workflow file. This is the rejig docs #1 recurring source of CI friction.
• Pre-commit validation failures: The two-checkpoint strategy is well-established in memory but still causing failures. Checkpoint 1: make build && make fmt after first major code edit. Checkpoint 2: make agent-report-progress (zero lint errors) before every PR. Enforce both.
• ANSI escape sequences in YAML: The validate-yaml job catches these before other jobs run — avoid terminal color codes leaking into YAML output.

Code Quality

• JavaScript file splitting: skills/javascript-refactoring/SKILL.md describes how to split large JS into separate .cjs files. Decision rule: files >300 lines or containing 2+ distinct domains should be split.
• Type patterns: Use distinct named types (e.g., JobName vs StepID) to prevent mixing bugs at compile time — well-documented in memory but worth reinforcing in reviews.
• Validation complexity: scratchpad/validation-refactoring.md has a step-by-step guide; teams working in validation code should consult it before adding more logic to existing validators.
• Code organisation: Prefer many smaller files grouped by functionality; add new files for new features rather than extending existing ones.

CI Health

• Stale lock files remain the top CI health concern (see Error Patterns above).
• Test suite runtime: make test takes >5 minutes — let CI handle it; use make test-unit (~3s) locally during development for fast feedback.
• Recommended local workflow: make build && make fmt → selective unit tests → push → let CI run the full suite.
• Workflow runbook: .github/aw/runbooks/workflow-health.md covers missing-tool errors, auth failures, MCP configuration issues, and safe-input/output problems — reference before raising issues.

Quick Wins

1. Add a git hook or Makefile guard that reminds developers to run make recompile when .aw/*.md files change — directly addresses the rejig docs #1 verified incident.
2. Enforce make fmt in a pre-push hook — eliminates the class of "5 CI failures in one day" pre-commit formatting mistakes.
3. Audit and prune 7 low-quality memories: Run hippo audit to identify and remove stale/misleading entries that may confuse future recall.
4. Reference skills/javascript-refactoring on any PR that touches large JS action files — the skill provides concrete splitting patterns.
5. Add make test-unit to the contribution guide as the recommended fast-feedback loop, reducing developer friction.

Longer-term Themes

• Two-checkpoint validation is a recurring pattern across many memories (8+ retrieves each). It's clearly important but still causing failures, suggesting it needs better tooling enforcement, not just documentation.
• Skills system adoption: The memory store repeatedly surfaces skill references (error-messages, javascript-refactoring, error-recovery-patterns, developer) — a periodic "skills health" review could ensure skills stay current and are being used.
• Semantic consolidation maturity: 531 active memories with 79 episodic→semantic merges per cycle suggests the store is healthy, but the 7 low-quality flags and 29 deduped entries indicate ongoing fragmentation worth addressing with a dedicated pruning pass.

Memory Health

• Total memories: 531 (high, healthy range)
• Decay removals this cycle: 0 (memories are fresh and strong)
• Duplicates deduped: 29 redundant semantic patterns removed
• Low-quality flags: 7 (run hippo audit list to review specifics)
• Auto-shared to global store: 174 high-value memories

Memory store stats from sleep cycle

Active memories:  531
Removed (decayed): 0
Merged episodic:   79
New semantic:      29
Deduped:           29 redundant semantic patterns
Auto-shared:       174 high-value memories
Embeddings:        531/531 (19 new this cycle)
Low-quality flags: 7 (run hippo audit)

References:

• §26020265916

Generated by 🦛 Daily Hippo Learn · ● 6.6M · ◷

• expires on May 21, 2026, 7:50 AM UTC

[github-actions[bot]]

This discussion has been marked as outdated by Daily Hippo Learn.

A newer discussion is available at Discussion #33492.