{"slug": "the-beaver-a-proposed-agent-skill-for-turning-recurring-human-approved-lessons", "title": "THE BEAVER: a proposed agent skill for turning recurring, human-approved lessons into structural repairs, so systems stop storing the same mistake and start preventing it.", "summary": "A developer proposed 'The Beaver,' an agent skill designed to convert recurring, human-approved lessons into structural repairs, preventing systems from repeatedly storing the same mistake without changing behavior. The skill outlines a loop for confirming failure recurrence, identifying the responsible code path, proposing minimal fixes, and maintaining human oversight before implementation.", "body_md": "| --- | |\n| name: the-beaver | |\n| description: Analyze repeated, human-approved lessons and draft the smallest structural repair that could prevent recurrence. Use when an agentic system keeps rediscovering the same failure, memory is accumulating without behavior changing, or a lesson may need to graduate into a hook, default, skill, template, or code path. Remain advisory unless accepted authority explicitly permits implementation. | |\n| --- | |\n| # The Beaver | |\n| ## Constraint | |\n| A stored lesson is not an improvement. If the same failure returns, the system has memory, but the dam still leaks. | |\n| ## Boundary | |\n| Work only from human-approved, provenance-backed lessons and current source. Treat semantic similarity as a lead, not proof of recurrence. | |\n| A human may waive the recurrence threshold on severity. One instance plus an explicit human call is sufficient authority for a proposal, never for automatic promotion. | |\n| Do not create new authority, promote a proposal, mutate production, or close a claim merely because a repair was suggested. If the system has no accepted rules for failure-class identity, recurrence thresholds, evidence anchors, subsumption, or the observation window, stop at a repair proposal and name the missing contract. | |\n| ## Loop | |\n| 1. **Collect the branch.** Read the approved lesson, its evidence anchors, and its provenance. | |\n| 2. **Confirm the leak.** Find independent instances of the same failure class using the system's accepted deterministic key. Do not infer recurrence from wording alone. | |\n| 3. **Find the owner.** Read the hook, default, skill, template, or code path that currently produces the behavior. | |\n| 4. **Shape the repair.** Propose the smallest change that could prevent the failure upstream. Reuse or amend the existing owner before adding a new layer. | |\n| 5. **Keep the human gate.** Present the evidence, proposed repair, denial behavior, and proof plan for approval. Never promote automatically. | |\n| 6. **Prove the structure.** Drive the repaired surface live and test the dangerous denial case. A landed change is not yet evidence that recurrence stopped. | |\n| 7. **Watch for silence.** Observe the accepted window. Only then report that the repair holds and let the lesson be subsumed by the structure that now carries it. | |\n| 8. **Reopen on return.** If the same failure class appears after subsumption, reopen the proposal. The failed repair is now evidence. Repair v2 starts from what v1 missed, never from scratch. | |\n| ## Output | |\n| Return: | |\n| - **Constraint:** the repeated failure in plain language. | |\n| - **Recurrence evidence:** independent, source-linked instances. | |\n| - **Existing owner:** the narrowest surface that should absorb the lesson. | |\n| - **Smallest repair:** the bounded structural change. | |\n| - **Dangerous denial:** what must still fail closed. | |\n| - **Human decision:** approve, edit, dismiss, or request more evidence. | |\n| - **Proof window:** how recurrence will be checked after the change. | |\n| - **Status:** `candidate`, `proposed`, `landed`, `observed`, `dismissed`, or `reopened`. | |\n| - **Retires:** the memory-only path or repeated workaround this repair makes unnecessary. | |\n| ## Standard | |\n| Do not call a system self-improving because it stored more memory. | |\n| Call the repair proven only when the structure changed and the same mistake stopped recurring. |", "url": "https://wpnews.pro/news/the-beaver-a-proposed-agent-skill-for-turning-recurring-human-approved-lessons", "canonical_source": "https://gist.github.com/firechickensolutions/0dd8e2908036c9ac95851212b80f8fda", "published_at": "2026-08-12 16:27:46+00:00", "updated_at": "2026-08-17 21:11:20.927943+00:00", "lang": "en", "topics": ["ai-agents", "ai-safety", "developer-tools"], "entities": [], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/the-beaver-a-proposed-agent-skill-for-turning-recurring-human-approved-lessons", "markdown": "https://wpnews.pro/news/the-beaver-a-proposed-agent-skill-for-turning-recurring-human-approved-lessons.md", "text": "https://wpnews.pro/news/the-beaver-a-proposed-agent-skill-for-turning-recurring-human-approved-lessons.txt", "jsonld": "https://wpnews.pro/news/the-beaver-a-proposed-agent-skill-for-turning-recurring-human-approved-lessons.jsonld"}}