🦛 Hippo Memory Insights — 2026-07-05

[github-actions[bot]]

Summary

• Memories in store: 342
• New lessons learned from commits today: 0
• Highest-confidence memory: recurring incident that stale workflow .lock.yml files create CI churn; run make recompile after markdown workflow edits

Top Memories Surfaced

• mem_4afaa0965161 — Recurring incident: stale workflow .lock.yml files cause churn and CI friction; run make recompile after markdown workflow edits.
• mem_bf15fc906323 — The validate-yaml job in .github/workflows/ci.yml scans YAML for ANSI escape sequences before other jobs run.
• mem_54c981bfc72e — skills/javascript-refactoring/SKILL.md is the preferred guide for splitting/refactoring JavaScript into .cjs files.
• sem_563940a28c81 — Consolidated lesson from error-reporting fixes: improve assignment error reports with correct titles and permissions guidance.
• mem_48a6fa8403bf — Outcome reporting and OTLP data quality validation has been a recent maintenance hotspot.
• mem_7fbf46669778 — Prefer many smaller files grouped by functionality.
• mem_1c38df66aa62 — Semantic types help prevent mixed-domain bugs such as using JobName where a StepID is expected.

Suggested Improvements

Error Patterns

• Strengthen pre-PR validation around workflow churn by making make recompile the default follow-up to edits under .github/workflows/*.md. This is the clearest recurring failure pattern in memory and already appears as an incident-level lesson.
• Continue improving error-report quality in workflow/assignment paths. Recent memories point to fixes around titles and permissions guidance, suggesting users still lose time on actionable-but-poorly-presented failures. Focus likely areas around CLI/workflow validation and assignment reporting paths.
• Add or expand tests that protect typed identifier boundaries in Go code, especially around workflow/job/step identifiers under pkg/. The memory store repeatedly highlights semantic types as a bug-prevention mechanism, which suggests this is a proven defense rather than a style preference.

Code Quality

• Prioritize refactors that split oversized or multi-purpose files, especially validators and command implementations. Memory consistently recommends many smaller files, a 100–200 line validator target, and avoiding growth-by-extension when adding features.
• Keep migrating dynamic configuration handling toward typed frontmatter fields in pkg/workflow/frontmatter_types.go and related codepaths. Several memories emphasize compile-time safety and clearer domain types.
• Review places still relying on broad interfaces or generic aliases where concrete domain types would improve safety. High-signal memories repeatedly point to WorkflowID, JobName, StepID, EngineName, and related semantic aliases as worthwhile guardrails.

CI Health

• The memory store strongly suggests that local selective testing plus CI-owned full-suite validation remains the right tradeoff. Preserve that pattern, but make fast local gates more visible: make build && make fmt, then make agent-report-progress.
• Keep the ANSI/YAML validation path in .github/workflows/ci.yml healthy and well-documented. Memories show ANSI contamination in generated or copied workflow YAML is a known recurring class of failure.
• Consider adding a lightweight check or authoring hint that flags deprecated workflow expressions and reminds authors to use steps.sanitized.outputs.* directly. The recurrence of those memories suggests avoidable compile/deprecation noise.

Quick Wins

• Add a focused doc or lint hint near workflow authoring docs that says: after editing .github/workflows/*.md, always run make recompile. This is the highest-confidence, highest-repeat operational improvement.
• Add targeted tests around typed workflow identifiers in pkg/ to catch cross-assignment mistakes early. This is small-scope and directly supported by multiple memories.
• Identify one or two large validators/command files and split them into smaller domain files following the existing validation refactoring guidance in scratchpad/validation-refactoring.md.
• Audit JavaScript workflow helper code and ensure any larger scripts follow the .cjs refactoring guidance in skills/javascript-refactoring/SKILL.md.
• Refresh contributor-facing docs to emphasize fast local validation versus slow full-suite runs, reducing wasted local time while preserving CI confidence.

Longer-term Themes

• Workflow authoring remains the dominant systems theme: recompilation, expression migrations, YAML hygiene, cache/artifact filename constraints, and CI validation all cluster together. This suggests a dedicated workflow-authoring UX pass would pay off.
• Type-safety as architecture keeps resurfacing. The memory store repeatedly favors semantic aliases, typed frontmatter, and smaller focused modules, indicating a broad long-term direction toward clearer boundaries in pkg/workflow/ and pkg/cli/.
• Operational reliability lessons are increasingly encoded as process memories rather than one-off fixes. Converting the highest-value memories into automated checks, tighter docs, or explicit guardrails would reduce repeated human rediscovery.

Memory Health

• Store health is generally good: 342 total memories, 342 embedded, average strength 0.81, and 0 open conflicts.
• There are 40 at-risk memories and 30 stale memories, so the next pruning pass should target low-signal imported fragments such as generic headings (for example memories that only contain phrases like "When to use each approach" or "Implication for code changes") because they rank but add little standalone value.
• The architectural-decisions recall surfaced several weak standalone fragments instead of concrete design decisions, which suggests that category would benefit from consolidation or curation into fewer, more self-contained semantic memories.

Supporting notes

• hippo learn --git found no fix/revert/bug commits in the last 7 days, so today's discussion is driven by the existing store rather than fresh commit-derived lessons.
• hippo sleep and hippo export hit intermittent database is locked failures during this run, but hippo status confirmed the store is healthy and fully embedded, and all four recall lenses completed successfully.
• Because hippo list is not available in this installation, memory-health counts were taken from hippo status instead.

Generated by 🦛 Daily Hippo Learn · 20.9 AIC · ⌖ 18.2 AIC · ⊞ 2.8K · ◷

• expires on Jul 7, 2026, 11:25 PM UTC-08:00

[pelikhan]

/q update hippo to create an issue with the hippo quick wins (3 wins max). add "cookie" label.

[github-actions[bot]]

🔧 Pay attention, 007! Q is preparing your gadgets for this discussion comment...

[github-actions[bot]]

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

A newer discussion is available at Discussion #44230.