🦛 Hippo Memory Insights — 2026-07-11

[github-actions[bot]]

Summary

• Memories in store: 340
• New lessons learned from commits today: 0
• Highest-confidence memory: mem_4afaa0965161 — stale workflow .lock.yml files cause churn and CI friction; 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_fae7a63e904c — Use the two-checkpoint validation strategy: make build && make fmt after the first major edit, then make agent-report-progress before report_progress.
• sem_ef67ccfd978b — make agent-report-progress is the fast pre-PR gate (build + fmt + lint + test-unit).
• mem_f7c45403b6e1 — New and updated workflow markdown should use steps.sanitized.outputs.*; the old needs.activation.outputs.* form is deprecated.
• mem_0324334270d0 — Guard list_code_scanning_alerts with state: open and severity: critical,high to avoid oversized MCP responses.
• mem_e02072c9ae07 — Self-hosted/GPU paths can fail with node: command not found; validate runtimes and PATH early.
• mem_ad0159793af4 — Stronger type patterns and semantic aliases prevent common string/int mix-up bugs.

Suggested Improvements

Error Patterns

• Reinforce the recurring workflow-churn fix around .github/workflows/*.md edits by making make recompile harder to forget. Memory mem_4afaa0965161 and related recompile memories show this is a repeated source of CI friction.
• Keep validation failures actionable by preserving the two-checkpoint routine from mem_fae7a63e904c: run make build && make fmt after the first major edit, and make agent-report-progress before shipping changes.
• For workflow prompts that call code-scanning helpers, always constrain list_code_scanning_alerts with state: open and severity: critical,high (mem_0324334270d0) so runs do not fail on oversized responses.

Code Quality

• Continue the migration toward stronger Go types in pkg/constants, pkg/cli, and pkg/workflow. The recurring type-pattern memories (mem_ad0159793af4, mem_1c38df66aa62, mem_3fd7d5dffe35) suggest real bug-prevention value from semantic aliases like StepID, JobName, and WorkflowID.
• Clean up workflow markdown to use steps.sanitized.outputs.text/title/body directly rather than the deprecated needs.activation.outputs.* form (mem_f7c45403b6e1, mem_f4f37acf5f61, mem_1daae554e017). This reduces warning noise and keeps examples current.

CI Health

• The strongest CI pattern is still “recompile after workflow markdown edits” (mem_4afaa0965161, mem_54f44b8a2960, mem_3c0154ff3f2d). A lightweight enforcement script or lint check would likely pay back quickly.
• The store repeatedly warns that immediate CI failures waste sessions (mem_72915af0db42, sem_1cfa35f0999d). Keeping make agent-report-progress visible in contributor docs and automation remains high leverage.
• Self-hosted runner reliability would benefit from an early runtime check for Node (mem_e02072c9ae07), especially in workflows that depend on JS-based tools.

Quick Wins

1. Add a lightweight guard that fails fast when workflow markdown changes are present but regenerated .lock.yml outputs are stale. Memory: mem_4afaa0965161, mem_54f44b8a2960, mem_3c0154ff3f2d.
2. Sweep workflow markdown examples and prompts to replace deprecated needs.activation.outputs.text/title/body usage with steps.sanitized.outputs.*. Memory: mem_f7c45403b6e1, mem_f4f37acf5f61, mem_1daae554e017.
3. Add an early validation/helper for workflows that call list_code_scanning_alerts so prompts consistently include state: open and severity: critical,high. Memory: mem_0324334270d0.
4. Add or document an early Node runtime/PATH preflight for self-hosted or GPU-oriented workflows to prevent node: command not found failures. Memory: mem_e02072c9ae07.
5. Identify one high-traffic package to adopt stronger semantic type aliases where raw strings are still mixed (pkg/cli / pkg/workflow). Memory: mem_ad0159793af4, mem_1c38df66aa62, mem_3fd7d5dffe35.

Longer-term Themes

• Validation discipline is a systemic theme: the memory store repeatedly surfaces the same build/fmt/lint/test gatekeeping habits, which suggests tooling should encode them rather than relying on human recall.
• Workflow authoring ergonomics remain a recurring source of incidents: recompilation drift, deprecated expression forms, and oversized tool responses all point to value in stronger linting and autofix support in the compiler/tooling.
• Type-safety guidance appears often enough to justify a focused cleanup pass across Go packages that still rely on loosely-typed strings and any.

Memory Health

• Current store health from hippo status: 340 total memories, 56 at risk (<0.2 strength), 0 open conflicts, average strength 0.81.
• Consolidation merged 11 episodic memories into 4 new semantic memories during today’s sleep cycle.
• Hippo reported 4 low-quality memories during consolidation; those are worth reviewing with hippo audit in a dedicated cleanup pass.
• I could not use the requested hippo list command because this Hippo build does not implement list; hippo status and hippo export provided the usable counts instead.

Operational detail

• learn --git found no fix/revert/bug commits in the last 7 days, so no new commit-derived lessons were added today.
• sleep completed successfully: 340 active memories, 11 episodic merges, 4 new semantic memories, 4 duplicates deduped, and 257 high-value memories auto-shared.
• embed reported full coverage: 340/340 memories embedded.
• Some hippo recall calls intermittently failed with database is locked; rerunning queries sequentially worked for the most important lenses.

Next Actions

• Triage the top three quick wins below into implementation issues.
• Consider a follow-up maintenance pass to review the 4 low-quality memories flagged by Hippo audit.
• Investigate the intermittent Hippo database is locked failures in the daily runner so recall analysis is more reliable.

Generated by 🦛 Daily Hippo Learn · 23.9 AIC · ⌖ 21.3 AIC · ⊞ 5.4K · ◷

• expires on Jul 13, 2026, 11:03 PM UTC-08:00

[github-actions[bot]]

Me test. You run.

Warning

Firewall blocked 6 domains

The following domains were blocked by the firewall during workflow execution:

• accounts.google.com
• android.clients.google.com
• clients2.google.com
• contentautofill.googleapis.com
• safebrowsingohttpgateway.googleapis.com
• www.google.com

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "accounts.google.com"
    - "android.clients.google.com"
    - "clients2.google.com"
    - "contentautofill.googleapis.com"
    - "safebrowsingohttpgateway.googleapis.com"
    - "www.google.com"

See Network Configuration for more information.

📰 BREAKING: Report filed by Smoke Copilot - AOAI (apikey) · 24.9 AIC · ⌖ 3.03 AIC · ⊞ 17.9K · ◷
_{Comment /smoke-copilot-aoai-apikey to run again
Add label smoke to run again}