Skip to content

hoavdc/CodexKit / skills/codexkit-skill-template

codexkit-skill-template

Verified

Reference implementation of a v2 CodexKit skill. Copy this folder as a starting point for new skills. Demonstrates all 10 required sections, 4C verification, and the tier-2 folder structure.

FileKindSize
CHANGELOG.mdmarkdown217 B
SKILL.mdskill md4.2 KB
agents/openai.yamlconfig238 B
examples/common-mistakes.mdmarkdown1.7 KB
examples/good-output.mdmarkdown1012 B
verification/checklist.mdmarkdown1.1 KB

examples/common-mistakes.md

View on GitHub
# Common Mistakes — Skill Template

> Anti-patterns frequently seen in skill authoring. Avoid these.

## Mistake 1: Vague Procedure Steps

❌ **Bad:** "Step 2 — Analyze the data"

✅ **Good:** "Step 2 — Segment responses by department. Produce a cross-tabulation table with N, Mean, and significance flag."

**Why it matters:** An AI agent cannot execute "analyze." It needs a concrete deliverable.

---

## Mistake 2: Generic Verification Questions

❌ **Bad:** "Is the output correct?"

✅ **Good:** "Does the NPS calculation exclude neutral responses (7–8) from both promoter and detractor counts?"

**Why it matters:** Generic questions produce generic answers. Domain-specific questions catch domain-specific errors.

---

## Mistake 3: Missing Edge Cases

❌ **Bad:** No edge cases section, or "N/A"

✅ **Good:** "If the client describes needs in vague terms, run a clarification round before proceeding. Ask: What does success look like in 90 days?"

**Why it matters:** Edge cases are where AI fails most. Documenting them prevents repeated failures.

---

## Mistake 4: Compound Skills

❌ **Bad:** `codexkit-analyze-data-build-dashboard-send-report`

✅ **Good:** Three separate skills: `codexkit-data-analyzer`, `codexkit-dashboard-designer`, `codexkit-report-sender`

**Why it matters:** One skill = one task. Compound skills confuse agent routing and reduce reuse.

---

## Mistake 5: No Quality Criteria

❌ **Bad:** "Output should be good quality"

✅ **Good:**
- "All action items have a named owner and a calendar date"
- "No section exceeds 200 words without a table or visual break"
- "Executive summary is readable in under 60 seconds"

**Why it matters:** Without criteria, "done" is subjective. With criteria, "done" is verifiable.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 467kUpdated Claude CodeCodexVerifiedInstall
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 467kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 319kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 319kUpdated Claude CodeCodexVerifiedInstall