skill-evolution

name	skill-evolution
description	Use when the user asks about skill upgrades, update recommendations, or an evolution backlog; analyze io-trace and conduct_feedback to produce report-only evolution candidates. No automatic edits.
compatibility	["Python 3.11+ standard library"]

skill-evolution

skill-evolution reads io-trace JSONL and produces evolution candidate reports for repeated tool, workflow, and sequence patterns. If a session-intent ledger is provided, it uses the intent summary to explain why the tool sequence occurred.

This skill is report-only. It does not edit skills, install hooks, promote memory, or create new automation.

Contents

• When To Use
• Conduct Feedback
• Procedure
• Output Format
• Quality Decisions
• Warnings

When To Use

• Analyze ~/.ghost-alice/io-trace.jsonl or isolated fixtures.
• Find repeated tool sequences in a project.
• Interpret repeated behavior with .tmp/session-intent/<platform>/<session-id>/intent-state.json.
• Answer when the user asks whether there are skill upgrades, update recommendations, or evolution backlog items to review.
• Gather evidence before proposing a new skill, hook, or doctor check.

Conduct Feedback

Tool-frequency, workflow, and sequence mining finds automation candidates, but it does not show whether the agent's behavior matched what the user asked. conduct_feedback fills that gap. Read it from the session-intent ledger as a first-class evolution signal.

Open conduct_feedback entries form the update-recommendation backlog. occurrence_count preserves repeated same-id corrections inside one session, so chronic completion or workflow failures do not collapse into one visible event. A source=user-explicit correction or a recurring pattern is high priority. Route each candidate through necessity-gate and name the gate skill that should have prevented it, such as boundary-contract for scope failures, using-coding-convention for execution discipline, or verification-before-completion for false completion.

When the user asks whether there are skill upgrades, update recommendations, or evolution backlog items, enumerate the open entries. For the full cross-session backlog, run scripts/aggregate_recommendations.py. For one ledger, run scripts/analyze_io_trace.py --intent-ledger <intent-state.json>, where open entries surface as conduct:<id> candidates even without tool activity. This skill stays report-only; accepted edits happen only under a separate explicit skill-update task.

Procedure

1. Choose the JSONL path and window size.
2. If available, add --intent-ledger <intent-state.json>.
3. Run scripts/analyze_io_trace.py <path> --json [--intent-ledger <intent-state.json>].
4. Read quality_summary first and compare review, watch, and reject.
5. Inspect the instincts field, which contains evolution candidates grouped by tool frequency, workflow, and sequence. Treat only quality=review candidates as possible follow-up work.
6. Use intent_context only as supporting explanation.
7. Route quality=review candidates through necessity-gate before action.
8. Treat quality=reject and decision=route-to-systematic-debugging as debugging or noise, not skill-evolution candidates.
9. Apply any accepted change in a separate plan. Do not change skills, hooks, or memory automatically while this skill runs; it stays report-only.

Output Format

{
  "window": 1000,
  "event_count": 3,
  "quality_summary": {
    "review": 1,
    "watch": 0,
    "reject": 0
  },
  "instincts": [
    {
      "id": "sequence:read-edit-bash",
      "trigger": "Read -> Edit -> Bash sequence",
      "confidence": 0.7,
      "domain": "sequence",
      "scope": "project",
      "evidence": "3 occurrences",
      "quality": "review",
      "decision": "necessity-gate",
      "session_count": 2,
      "quality_reasons": ["cross-session-evidence"],
      "intent_context": {
        "current_goal": "session intent guard implementation",
        "constraints": ["do not store raw prompts"],
        "decision_count": 1
      }
    }
  ]
}

Quality Decisions

• review: Cross-session evidence suggests the pattern may generalize. Do not implement directly; route through necessity-gate.
• watch: Single-session or low-frequency evidence. Keep observing.
• reject: The pattern is a local test/debug loop or otherwise not a reusable skill evolution candidate. If decision=route-to-systematic-debugging, hand it off to the systematic-debugging loop.

Warnings

• Default mode is report-only.
• Do not auto-evolve, auto-promote, auto-create skills, or add background hooks.
• Do not depend on personal absolute paths such as /Users/aidall.
• intent_context must come from session-intent-analyzer compressed state, not raw prompts.
• confidence is a prioritization signal, not approval evidence.
• quality=review is a review entry condition, not approval.
• Do not use external network, MCP, or credential surfaces.