skill-evolve
- Repo stars 479
- Author updated Live
- Author repo UniClipboard
- Domain
- Data
- Compatible agents
-
- Claude Code
- Cursor
- Cline
- Codex
- Windsurf
- Gemini CLI
- +20
- Trust score
- 88 / 100 · community maintained
- Author / version / license
- @UniClipboard · no license declared
- Token usage
- Lean
- Setup complexity
- Guided setup
- External API key
- Not required
- Operating systems
- Unspecified (assume cross-platform)
- Runtime requirements
- No special requirements
- Permissions
-
- Read-only
- Write / modify
- Shell exec
- Network behavior
- Local-only
- Install commands
- 26 variants
Profile is derived at build time from SKILL.md and install vectors. Subject to drift from author intent.
Heads up: 未限定 allowed-tools，默认拥有全部工具权限。
---
name: skill-evolve
description: Meta-skill for evolving other skills. Invoke after resolving a real-world issue to extract lesso…
category: data
runtime: no special runtime
---
# skill-evolve output preview
## PART A: Task fit
- Use case: Meta-skill for evolving other skills. Invoke after resolving a real-world issue to extract lessons and merge them into the referenced SKILL.md. Usage - /skill-evolve @path/to/SKILL.md.
- Inputs: target material, constraints, expected output, and acceptance criteria.
- Evidence boundary: follow “When to Use / Procedure / Step 1 — Gather Context” and do not present inference as author intent.
## PART B: Execution result
- **01** The card summarizes the use case; runtime output centers on “Meta-skill for evolving other skills. Invoke after resolving a real-world issue to extract lessons and merge them into the referenced SKILL.md. Usage - /skill-evolve @path/to/SKILL.md”.
- **02** When the source has headings, the agent prioritizes “When to Use / Procedure / Step 1 — Gather Context” so the result follows the author’s structure.
- **03** Typical output includes task judgment, concrete steps, required commands or file edits, validation, and follow-up options.
- **04** Risk context follows the fingerprint: read files, write/modify files, run shell commands; mostly runs locally; usually needs no extra API key.
## Running Rules
- read files, write/modify files, run shell commands; mostly runs locally; usually needs no extra API key.
- Validate with a small sample before expanding scope.
- Return the result, validation criteria, and next iteration options. Choose the trigger
The source mentions slash commands such as `/skill-evolve`; use them first when your agent supports command triggers.
Provide inputs and boundaries
Name target files or source material, expected output, forbidden changes, and whether network or shell access is allowed. Permission fingerprint: read files, write/modify files, run shell commands.
Validate with a small run
Start with a small task and check whether the result follows “When to Use / Procedure / Step 1 — Gather Context”. Inspect diffs, logs, previews, or tests before expanding scope.
Review before delivery
Confirm the final output includes a concrete result, evidence, and next action. If it stays generic, tighten inputs, boundaries, and acceptance criteria.
---
name: skill-evolve
description: Meta-skill for evolving other skills. Invoke after resolving a real-world issue to extract lesso…
category: data
source: UniClipboard/UniClipboard
---
# skill-evolve
## When to use
- Meta-skill for evolving other skills. Invoke after resolving a real-world issue to extract lessons and merge them into…
- Use it when the task has clear inputs, repeatable steps, and validation criteria.
## What to provide
- Target material, scope, expected result, and forbidden changes.
- Whether network, commands, file writes, or external services are allowed.
## Execution rules
- Organize steps around “When to Use / Procedure / Step 1 — Gather Context” and keep inference separate from source facts.
- read files, write/modify files, run shell commands; mostly runs locally; usually needs no extra API key.
- Validate with a small sample before expanding the task.
## Output requirements
- Return the deliverable, key evidence, validation method, and next action.
- Mark missing information as unknown; do not invent commands, platforms, or dependencies. Evidence boundary and execution chain
The author source anchors workflow facts; repository files anchor sources and commands; Fluxly only adds fit, limitations, and quality judgment.
skill "skill-evolve" {
input -> user goal + target files + boundaries + acceptance criteria
context -> When to Use / Procedure / Step 1 — Gather Context
rules -> SKILL.md triggers / order / output contract
runtime -> no special runtime | read files, write/modify files, run shell commands | mostly runs locally
guardrails -> usually needs no extra API key + small-sample validation + diff/log review
output -> copyable result + checklist + next iteration
}Skill Evolve
Turn real-world experience into durable skill knowledge.
When to Use
User invokes this skill after completing work that revealed a gap, pitfall, or new pattern not yet captured in an existing SKILL.md. Typical trigger: /skill-evolve @path/to/SKILL.md
Procedure
Step 1 — Gather Context
- Read the referenced SKILL.md in full.
- Review the current conversation to identify:
- What went wrong or was discovered (the incident/finding)
- Root cause (why it happened)
- The fix or pattern (what the correct approach is)
- Why the existing skill didn't prevent it (the gap)
Step 2 — Classify the Learning
Determine which type of update is needed:
| Type | Description | Example |
|---|---|---|
| New Rule | A pattern/anti-pattern not covered at all | .entered() in async is forbidden |
| Rule Refinement | Existing rule needs caveats or edge cases | skip_all needed when params include trait objects |
| New Example | Correct/incorrect code pattern to illustrate existing rule | async {}.instrument(span).await for match arms |
| Checklist Addition | New review checkpoint | ".entered() held across .await? FORBIDDEN" |
| Section Expansion | Existing section needs a new subsection | Adding async lifecycle patterns to spawn section |
Step 3 — Draft the Update
Write the proposed additions following these principles:
- Match the existing style — same heading levels, table formats, code block conventions
- Place it where it belongs — near related content, not appended randomly at the end
- Lead with the rule, then the why, then the example
- Include both CORRECT and FORBIDDEN patterns when adding anti-patterns
- Keep it concise — skills are reference material, not tutorials
Step 4 — Apply and Summarize
- Edit the SKILL.md with the new content.
- Present a summary to the user:
## Skill Evolution Summary
**Skill**: <skill name>
**Trigger**: <what happened that revealed the gap>
**Update type**: <New Rule | Rule Refinement | New Example | Checklist Addition | Section Expansion>
**What was added**:
- <bullet summary of each change>
**Section(s) modified**: <list of modified section numbers/names>Quality Gates
Before applying the update, verify:
- The new content does NOT duplicate existing rules (check the full SKILL first)
- The new content is grounded in a real incident from this conversation (not hypothetical)
- Code examples compile conceptually (correct syntax, realistic types)
- The learning is generalizable — useful beyond this one specific case
- If adding a FORBIDDEN pattern, a CORRECT alternative is shown alongside it
Anti-Patterns for This Skill
- Don't bloat: If the learning is too narrow or one-off, suggest the user add it to memory instead of a skill
- Don't rewrite: Evolve the skill incrementally. Don't restructure or rewrite existing sections unless they are wrong
- Don't speculate: Only add patterns that were validated in practice during this conversation
- Don't duplicate: If the fix was specific to one codebase and not transferable, it belongs in CLAUDE.md, not a skill
Decide Fit First
Design Intent
How To Use It
Boundaries And Review