ai-md
Convert human-written CLAUDE.md into AI-native structured-label format. Battle-tested across 4 models. Same rules, fewer tokens, higher compliance.
36
33%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Risky
Do not use without reviewing
Optimize this skill with Tessl
npx tessl skill review --optimize ./plugins/antigravity-awesome-skills-claude/skills/ai-md/SKILL.mdQuality
Discovery
40%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
The description identifies a clear and distinctive niche (CLAUDE.md conversion) but suffers from marketing language ('battle-tested', 'higher compliance') that adds no selection value. It completely lacks explicit trigger guidance ('Use when...') and misses natural keyword variations that users might employ when needing this skill.
Suggestions
Add an explicit 'Use when...' clause, e.g., 'Use when the user wants to reformat, compress, or optimize their CLAUDE.md file, or mentions reducing token usage in system instructions.'
Replace marketing fluff ('battle-tested across 4 models', 'higher compliance') with concrete actions like 'Parses human-written rules, converts prose to labeled directives, reduces token count while preserving instruction semantics.'
Include natural trigger terms users might say, such as 'optimize CLAUDE.md', 'reformat instructions', 'reduce tokens', 'compress system prompt', or 'structured labels'.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Names the domain (CLAUDE.md conversion) and one specific action (convert to structured-label format), but 'battle-tested across 4 models' and 'same rules, fewer tokens, higher compliance' are marketing fluff rather than concrete actions. Does not list multiple specific capabilities. | 2 / 3 |
Completeness | Describes what it does (convert CLAUDE.md to structured-label format) but completely lacks a 'Use when...' clause or any explicit trigger guidance for when Claude should select this skill. Per rubric guidelines, missing 'Use when' caps completeness at 2, and the 'when' is so absent this warrants a 1. | 1 / 3 |
Trigger Term Quality | Includes 'CLAUDE.md' and 'structured-label format' which are relevant keywords, but misses natural variations users might say like 'reformat instructions', 'optimize system prompt', 'compress CLAUDE.md', or 'token reduction'. The phrase 'AI-native' is jargon rather than a natural user term. | 2 / 3 |
Distinctiveness Conflict Risk | The skill targets a very specific niche—converting CLAUDE.md files to a particular structured-label format. This is unlikely to conflict with other skills given the highly specific input (CLAUDE.md) and output (AI-native structured-label format). | 3 / 3 |
Total | 8 / 12 Passed |
Implementation
27%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill contains genuinely valuable methodology for converting prose instructions to structured AI-native format, with good concrete examples of before/after transformations. However, it severely violates its own core principle — it's extremely verbose prose explaining why brevity matters, spending hundreds of tokens on theory (attention mechanisms, semantic anchoring) that Claude doesn't need to perform the conversion. The irony of a token-efficiency skill being this token-inefficient significantly undermines its effectiveness.
Suggestions
Cut the entire 'Why It Works' section (Mechanisms 1-3) — Claude doesn't need to understand attention theory to perform conversions. Move it to a separate THEORY.md if users want background.
Split into multiple files: SKILL.md (overview + quick start + template), CONVERSION-PHASES.md (the 6-phase process), TESTING.md (the exam protocol), REFERENCE.md (label vocabulary table + anti-patterns).
Replace the descriptive Phase 6 testing section with a concrete, executable test script or checklist with explicit pass/fail thresholds (e.g., 'If any model scores below 7/8, revert and re-examine the specific failing rule').
Remove the 'Real-World Results' narrative and condense to just the comparison table with a one-line takeaway — the editorial commentary ('The uncomfortable truth...') is exactly the kind of human motivation text the skill itself says to delete.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose at ~400+ lines. Extensively explains HOW and WHY LLMs process instructions (attention splitting, semantic anchoring) — concepts that are explanatory background, not actionable instructions. The 'Why It Works' section alone is a lengthy tutorial that Claude doesn't need. Massive amounts of prose explaining the methodology rather than just providing the conversion steps. | 1 / 3 |
Actionability | The conversion phases (1-6) provide a structured process with concrete examples of input→output transformations, label vocabularies, and a template. However, much of it is descriptive rather than executable — there are no runnable scripts for the actual conversion, the bash snippet is just a token counter, and the 'test protocol' describes what to do conceptually rather than providing executable test harnesses. | 2 / 3 |
Workflow Clarity | The six-phase conversion process (Understand→Decompose→Label→Structure→Resolve→Test) is clearly sequenced, and the two-stage workflow (Preview→Distill) includes a backup step. However, validation checkpoints are weak — Phase 6 testing is described conceptually but lacks concrete pass/fail criteria or feedback loops for what to do when specific phases fail. The 'ask user before proceeding' is good but there's no explicit validation between phases 1-5. | 2 / 3 |
Progressive Disclosure | Everything is in a single monolithic file with no bundle files or external references. The content is extremely long and would benefit enormously from splitting — the 'Why It Works' theory section, the label vocabulary reference, the template, the anti-patterns table, and the test protocol could all be separate referenced files. As-is, it's a wall of text that loads hundreds of lines of explanatory content every time. | 1 / 3 |
Total | 6 / 12 Passed |
Validation
81%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 9 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
skill_md_line_count | SKILL.md is long (524 lines); consider splitting into references/ and linking | Warning |
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
Total | 9 / 11 Passed | |
- Repository
- sickn33/antigravity-awesome-skills
- Commit
a4d3e3a
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.