Engineering AgentSkillsExtraction

Self-Improving — Extract

Extract recurring solutions into reusable skills.

$skillfish add ~skills/self-improving-agent-extract

Real Examples Get Started Models & Context

Real Conversations

See how the Self-Improving — Extract skill responds to actual requests. These are representative examples of what your AI agent will produce.

You

I keep hitting the same Docker platform mismatch error on my M2 Mac across multiple projects: WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8). Can you extract this into a reusable skill so I stop debugging it from scratch?

Agent

Perfect extraction candidate — non-obvious, recurring across projects, multi-step fix. Creating docker-m1-fixes.

Searching MEMORY.md... Found 3 related entries across 2 sessions.

Generated: docker-m1-fixes/SKILL.md

---
name: docker-m1-fixes
description: Fix Docker platform mismatch on Apple Silicon (M1/M2). Use when: seeing linux/amd64 vs linux/arm64 warning or containers fail to start on Mac.
---

## Quick Reference
| Problem | Solution |
|---|---|
| platform mismatch warning | Add `--platform linux/amd64` to build |
| Container fails to start on M1/M2 | Set `DOCKER_DEFAULT_PLATFORM=linux/amd64` |
| Slow builds via Rosetta | Use native `linux/arm64` image if available |

## Solutions

### Option 1: Build flag (one-time)
```bash
docker build --platform linux/amd64 -t myapp .

Option 2: docker-compose.yml (persistent)

services:
  app:
    platform: linux/amd64

Option 3: Global env var

export DOCKER_DEFAULT_PLATFORM=linux/amd64  # add to .zshrc


**Files created:**

docker-m1-fixes/SKILL.md (47 lines) docker-m1-fixes/README.md (18 lines) docker-m1-fixes/reference/examples.md (32 lines)


**Install anywhere:** `skillfish add ./docker-m1-fixes`

MEMORY.md entries retained — the skill is portable, the memory stays project-specific. Publish to clawhub?

You

Every time someone edits our OpenAPI spec and forgets to regenerate the TypeScript client, we get type errors. I've explained this 4 times in code review. Extract this into a skill.

Agent

Classic workflow amnesia — extract once, reference everywhere. Creating api-client-regen.

MEMORY.md search: Found 2 related entries — "TypeScript client out of sync with spec" and the openapi-generator-cli command.

Generated: api-client-regen/SKILL.md

---
name: api-client-regen
description: Regenerate TypeScript API client after OpenAPI spec changes. Use when: editing any file in openapi/ or src/api/spec/.
---

# API Client Regeneration

> Spec changed? Regenerate before committing.

## The Command
```bash
pnpm run generate:api-client
# or directly:
npx openapi-generator-cli generate \
  -i openapi/spec.yaml \
  -g typescript-axios \
  -o src/api/generated/

CI Enforcement

Fail the build if generated files are stale:

- name: Verify API client is up to date
  run: |
    pnpm run generate:api-client
    git diff --exit-code src/api/generated/

Common Failure Mode

Property 'userId' does not exist on type 'UserResponse' Spec added/renamed a field. Run regen.


**Also suggested:** Promote this to `.claude/rules/api.md` scoped to `openapi/**` — Claude will remind you proactively when editing spec files.

Create the scoped rule too?

Get Started

How to install and use this skill in your preferred environment.

Skills are designed for AI coding agents (Claude Code, Cursor, Windsurf) and IDE-based workflows where the agent can read files, run scripts, and act on your codebase. Web-based AI can use the knowledge and frameworks, but won't have tool access.

Models & Context

Which AI models and context windows work best with this skill.

Recommended Models

Best

Claude Opus 4 Claude Sonnet 4 GPT-4.1 Gemini 2.5 Pro Grok 3 Kimi K2

Good

Claude Haiku 4.5 GPT-4.1 mini Gemini 2.5 Flash Grok 3 mini

Larger models produce more detailed, production-ready outputs.

Context Window

This skill's SKILL.md is typically 3–10 KB — fits in any modern context window.

8K Skill only

32K+ Skill + conversation

100K+ Skill + references + codebase

All current frontier models (Claude, GPT, Gemini) support 100K+ context. Use the full window for complex multi-service work.

Pro tips for best results

1

Be specific

Include numbers — users, budget, RPS — so the skill can size the architecture.

2

Share constraints

Compliance needs, team size, and existing stack all improve the output.

3

Iterate

Start with a high-level design, then ask follow-ups for IaC, cost analysis, or security review.

4

Combine skills

Pair with companion skills below for end-to-end coverage.

Ready to try Self-Improving — Extract?

Install the skill and start getting expert-level guidance in your workflow — any agent, any IDE.