Agent Skills (SKILL.md) builder, auditor, and improver for cross-platform LLM agents.
Use for "skeall", "build a skill", "create skill", "improve skill", "audit skill",
"skill review", or any SKILL.md question. Follows agentskills.io standard.
Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
Agent Skills (SKILL.md) builder, auditor, and improver for cross-platform LLM agents.
Use for "skeall", "build a skill", "create skill", "improve skill", "audit skill",
"skill review", or any SKILL.md question. Follows agentskills.io standard.
Skeall
Create, improve, and audit Agent Skills following the Agent Skills open standard. This skill encodes lessons from real-world skill development and cross-platform compatibility testing.
Quick start
/skeall --create # Interview, then scaffold new skill
/skeall --improve <path> # Analyze and improve existing skill
/skeall --scan <path> # Audit only, no changes (report)
/skeall --scan . # Audit skill in current directory
/skeall --scan-all # Batch scan all skills in ~/.claude/skills/
/skeall --scan-all <dir> # Batch scan all skills in custom directory
/skeall --healthcheck <path> # Runtime check single skill (orphans, deps, env, URLs)
/skeall --healthcheck-all # Runtime check all skills in ~/.openclaw/skills/
/skeall --healthcheck-all <dir> # Runtime check all skills in custom directory
Mode 1: Create (scaffold a new skill)
Process
Interview the user (ask questions 1-4 always, then 5-6 if user hasn't already specified complexity or distribution scope):
YAML frontmatter with name and description (see Frontmatter section)
Body under 500 lines, under 5000 tokens
Instruction-based framing, not persona-based
Progressive disclosure: core in SKILL.md, details in references/
Show the generated SKILL.md to user for review.
Run --scan on the generated skill. If any HIGH issues found, fix them before delivering.
Next step: "Optimize with reprompter?" (optional, see Reprompter section). Then suggest installing the skill.
Mode 2: Improve (refactor existing skill)
Process
Read SKILL.md first. Read reference files only if scan identifies issues requiring them (broken links, routing table mismatches).
Run the scan checklist (see Mode 3).
For each issue found, propose a specific before/after edit.
Group edits by priority: HIGH first, then MEDIUM, then LOW.
Ask user: "Fix all? Review one by one? Or just the HIGHs?" (recommended: fix all HIGHs automatically, review MEDIUMs)
Apply approved edits.
Re-scan once. If new issues appear, report them but do not enter an infinite fix loop.
Next step: "Run --scan to verify?" or "Commit changes?"
Common improvements
Problem
Fix
Body over 5000 tokens
Move detail sections to references/
Redundant content
Single source of truth, reference elsewhere
Persona-based framing
Switch to instruction-based framing
Missing trigger phrases
Add keywords to description field
Platform-specific patterns
Replace with universal formatting
No progressive disclosure
Add routing table to reference files
Mode 3: Scan (audit and report)
Process
Read the skill's SKILL.md and directory structure.
Check every item in the checklist below.
Output a severity-tagged report.
Report format
## Skill Audit: {skill-name}
Score: X.X/10
STRUCTURE
[PASS] S1 -- SKILL.md exists at root
[FAIL] S3 HIGH -- name does not match directory name
[WARN] S5 MEDIUM -- No references/ directory
FRONTMATTER
[PASS] F2 -- Trigger phrases present
[FAIL] F1 HIGH -- description over 1024 characters
CONTENT
[WARN] C5 MEDIUM -- Persona-based framing ("You are an expert")
[FAIL] C3 HIGH -- Same content repeated 3 times (lines 45, 120, 280)
LLM-FRIENDLINESS
[WARN] L4 MEDIUM -- Unicode arrows instead of markdown tables
[PASS] L3 -- No emoji markers in headings
SECURITY
[PASS] SEC1 -- No XML angle brackets in frontmatter
[PASS] SEC3 -- No hardcoded secrets
CROSS-PLATFORM
[PASS] X1 -- No {baseDir} placeholders
[WARN] X4 LOW -- No multi-platform install instructions in README
Total: 3 HIGH | 4 MEDIUM | 1 LOW
Next step after scan: "Want me to fix these? Run /skeall --improve <path>"
Error handling
Input
Response
No SKILL.md found at path
"No skill found at {path}. Did you mean --create?"
Empty directory for --scan-all
"No skills found in {dir}. Skills must have a SKILL.md file."
Invalid YAML frontmatter
Report the parse error, suggest fixing frontmatter first
--improve on non-skill file
"Not a valid skill (no YAML frontmatter). Try --create instead."
--improve on a skill scoring 10/10
"Scan found 0 issues (score 10.0/10). No changes needed. Consider running trigger and functional tests."
Lowercase alphanumeric with hyphens only (unicode lowercase allowed)
1-64 characters, no leading/trailing/consecutive hyphens
No spaces, no special characters, no reserved words ("anthropic", "claude")
Recommended: gerund form (processing-pdfs, testing-code) or descriptive noun (pdf-processor)
description rules:
Explain WHAT it does AND WHEN to use it
Write in third person ("Processes files", not "I can process" or "You can use")
Include trigger phrases users would actually type
Put the most important keyword first (platforms weight first words)
Spec limit: 1024 characters. Recommended: under 300 for best matching
Use noun-phrase style ("Guide for X"), not persona style ("Expert in X")
No XML angle brackets (<, >) in any frontmatter value (injection risk)
Optional frontmatter fields
These are silently ignored by platforms that do not support them:
license:MIT# For distributed skillscompatibility:"Node.js 18+"# Environment requirements (max 500 chars)metadata:# Arbitrary key-value (author, version)author:your-nameversion:1.0.0allowed-tools:"Bash Read"# Experimental: space-delimited tool listuser-invocable:true# Show in /slash menu (false = hidden but still callable)disable-model-invocation:true# Block Claude from auto-loading this skillargument-hint:"<file-path>"# Hint shown in /skill autocompletemodel:opus# Override model for this skillcontext:fork# Run in isolated subagentagent:general-purpose# Subagent type: general-purpose, Explore, Plan, or customhooks:# Skill-scoped lifecycle hooksPostToolCall:"validate.sh"
Estimation: ~1.5 tokens per word for mixed code+prose markdown.
Progressive disclosure: SKILL.md body should handle ~70% of user requests. Reference files handle the remaining 30% (detailed workflows, complete examples, edge cases).
Line limits
Guideline
Limit
SKILL.md body
Under 500 lines (under 300 for complex skills with many references)
Reference files
No hard limit, but keep each under 700 lines. Add TOC at top if over 100 lines
Scan checklist
Structure checks
ID
Severity
Check
S1
HIGH
SKILL.md exists at skill root
S2
HIGH
YAML frontmatter present with --- delimiters
S3
HIGH
name field present and valid (lowercase, hyphens, 1-64 chars, no consecutive hyphens)
S4
HIGH
description field present
S5
MEDIUM
References in references/ not loose at root
S6
LOW
README.md present for GitHub-hosted skills
S7
LOW
No unnecessary files (node_modules, .DS_Store, etc.)
S8
HIGH
name field matches parent directory name
Frontmatter checks
ID
Severity
Check
F1
HIGH
Description under 1024 characters (spec limit)
F1b
LOW
Description under 300 characters (recommended for matching)
F2
HIGH
Description includes trigger phrases
F3
MEDIUM
Description starts with noun phrase, not "Expert in"
F4
MEDIUM
Name 1-64 characters, no leading/trailing/consecutive hyphens
F5
LOW
No platform-specific fields (keeps universal compatibility)
Content checks
ID
Severity
Check
C1
HIGH
Body under 500 lines
C2
HIGH
Estimated tokens under 5000
C3
HIGH
No content repeated in SKILL.md body (controlled repetition across reference files is acceptable)
C4
HIGH
Code examples use correct, verified patterns
C5
MEDIUM
Instruction-based framing (not "You are an expert")
C6
MEDIUM
Has routing table to reference files (if references/ exists)
C7
MEDIUM
Troubleshooting section present (for skills with code blocks or CLI commands)
C8
LOW
No deprecated content at the top (wastes prime token space)
C9
MEDIUM
Routing table completeness: if references/ exists, SKILL.md lists ALL files in references/
C10
MEDIUM
Internal count consistency: claimed counts ("34 patterns", "8 phases") match actual content
C11
MEDIUM
No stale references: documented APIs, functions, model names exist in actual source
LLM-friendliness checks
ID
Severity
Check
L1
HIGH
Tables for structured data (not bullet lists with arrows)
L2
HIGH
Imperative instructions ("Do X", not "You should consider X")
L3
MEDIUM
No emoji in headings or structural markers (frontmatter metadata values are data, not markers)
L4
MEDIUM
No Unicode arrows or special characters for data flow
Next step: "Start with the lowest-scoring skill. Run /skeall --improve <path> on it."
Mode 5: Health check (runtime audit)
Checks whether a skill actually works at runtime — beyond what static scan can catch. Run static scan (Mode 3) first and fix HIGH issues before health check.
Process
Run R1-R7 checks against the target skill.
For --healthcheck-all: cross-check all skills for duplicates (R2) and trigger collisions (R3).
Output severity-tagged report with sections: RUNTIME, DUPLICATES, TRIGGER COLLISIONS.
Labels: [FAIL] for HIGH issues (must fix), [WARN] for MEDIUM (runtime risk), [INFO] for LOW.
For detection algorithms, report format examples, and batch output format, see references/healthcheck.md.
Scoring methodology
Formula:Score = max(0, 10 - (HIGHs x 1.5) - min(MEDIUMs x 0.5, 3) - min(LOWs x 0.2, 1))
PASS threshold: Score 7+ AND zero HIGH issues. For detailed examples, see references/scoring.md.
Troubleshooting
Issue
Fix
Token estimate seems wrong
Use wc -w and multiply by 1.5 (prose) or 1.7 (code-heavy)
Scan reports FAIL but skill works fine
HIGHs indicate spec/LLM issues, not runtime bugs. Fix them anyway.
Batch scan misses a skill
Skill directory must contain SKILL.md at root
Two fixes contradict each other
Flag the conflict, ask user to choose (e.g., "shorten file" vs "add section")
Score 7+ but still NEEDS WORK
Check for HIGH issues. Any HIGH = NEEDS WORK regardless of score
Testing your skill: After create or improve, test trigger activation (3-5 keyword variants), functional output, and negative (unrelated queries stay quiet). See references/testing.md.
MCP integration: Use fully qualified tool names (mcp__server__tool_name). Document required MCP servers and provide fallbacks. See references/advanced-patterns.md.
Reprompter integration (optional): After --create interview, say "reprompter optimize" to score description variants and validate code examples. Works standalone if reprompter is not installed.
