Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Diagnose and fix agent behavioral surfaces when the user corrects a mistake — connects to Claude native memory.
.claude/skills/automagik-dev-learn/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 97% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 288% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 118% | 0% |
When the user corrects a mistake, diagnose which behavioral surface caused it and apply one minimal, targeted fix. Interactive — runs in the foreground with the user.
.claude/memory/MEMORY.md index so the learning survives across sessions.| Surface | Path | Controls | |---------|------|----------| | Project conventions | CLAUDE.md | Commands, gotchas, coding rules | | Agent identity | AGENTS.md | Role, preferences, team behavior | | Personality | SOUL.md / IDENTITY.md | Tone, communication style | | Global rules | ~/.claude/rules/*.md | Cross-project behavior | | Native memory | .claude/memory/ | Feedback, user prefs, project context | | Project memory | memory/ | Project-scoped knowledge | | Hooks | .claude/settings.json | Event-driven automation, gates |
skills/ and agents/.claude/memory/<name>.md:
markdown--- name: <concise-name> description: <one line for relevance matching> type: feedback --- <The rule itself> **Why:** <the incident or reason the user gave> **How to apply:** <when/where this guidance kicks in>
"Stop using pip install — this system only has uv." → Diagnosis: no Python-tooling rule exists anywhere; surface is global rules, since it applies to every project. Proposal in plan mode: create ~/.claude/rules/python-tooling.md (never pip; uv tool install for CLIs, uv pip install inside venvs). On approval: write the rule, then persist a use-uv-not-pip feedback memory and index it.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→pass | 7,559 | 5,222 | -31% | 1 | 1 | 0% | 1,253 | 1,557 | +24% | 0 | 0 | — |
case-10 | fail→pass | 12,437 | 9,073 | -27% | 1 | 1 | 0% | 2,158 | 2,152 | -0% | 0 | 0 | — |
case-01 | fail→fail | 4,708 | 6,337 | +35% | 1 | 1 | 0% | 794 | 1,005 | +27% | 0 | 0 | — |
case-02 | fail→pass | 5,742 | 6,173 | +8% | 1 | 1 | 0% | 861 | 1,698 | +97% | 0 | 0 | — |
case-03 | fail→fail | 8,823 | 4,825 | -45% | 1 | 1 | 0% | 1,342 | 982 | -27% | 0 | 0 | — |
case-05 | fail→fail | 9,452 | 16,948 | +79% | 1 | 1 | 0% | 1,472 | 973 | -34% | 0 | 0 | — |
case-06 | fail→fail | 8,578 | 5,914 | -31% | 1 | 1 | 0% | 1,441 | 1,072 | -26% | 0 | 0 | — |
case-07 | fail→fail | 5,130 | 2,022 | -61% | 1 | 1 | 0% | 790 | 943 | +19% | 0 | 0 | — |
case-08 | fail→pass | 2,767 | 5,399 | +95% | 1 | 1 | 0% | 381 | 1,477 | +288% | 0 | 0 | — |
case-09 | fail→pass | 5,555 | 8,227 | +48% | 1 | 1 | 0% | 955 | 2,086 | +118% | 0 | 0 | — |
case-11 | fail→fail | 4,610 | 3,394 | -26% | 1 | 1 | 0% | 647 | 1,098 | +70% | 0 | 0 | — |
case-12 | fail→pass | 10,159 | 3,655 | -64% | 1 | 1 | 0% | 1,653 | 1,259 | -24% | 0 | 0 | — |
case-13 | fail→pass | 8,249 | 3,388 | -59% | 1 | 1 | 0% | 1,235 | 1,204 | -3% | 0 | 0 | — |
case-14 | fail→pass | 9,101 | 6,219 | -32% | 1 | 1 | 0% | 1,498 | 1,738 | +16% | 0 | 0 | — |
case-15 | fail→pass | 5,369 | 6,319 | +18% | 1 | 1 | 0% | 734 | 1,644 | +124% | 0 | 0 | — |
case-16 | fail→pass | 4,266 | 4,855 | +14% | 1 | 1 | 0% | 584 | 1,406 | +141% | 0 | 0 | — |
case-17 | pass→pass | 8,307 | 4,481 | -46% | 1 | 1 | 0% | 1,416 | 1,382 | -2% | 0 | 0 | — |
case-18 | fail→pass | 8,524 | 4,205 | -51% | 1 | 1 | 0% | 1,457 | 1,244 | -15% | 0 | 0 | — |
case-19 | fail→pass | 4,544 | 19,554 | +330% | 1 | 1 | 0% | 203 | 3,377 | +1564% | 0 | 0 | — |
case-20 | fail→fail | 7,035 | 4,329 | -38% | 1 | 1 | 0% | 1,112 | 936 | -16% | 0 | 0 | — |
case-21 | pass→fail | 6,742 | 4,921 | -27% | 1 | 1 | 0% | 1,208 | 956 | -21% | 0 | 0 | — |
case-22 | pass→fail | 8,171 | 5,534 | -32% | 1 | 1 | 0% | 1,378 | 946 | -31% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 17 counted toward the lift figure. The other 5 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +45 percentage points is the difference between those two pass rates over the 17 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.