CtrlK
BlogDocsLog inGet started
Tessl Logo

axiom-audit-foundation-models

Use when the user mentions Foundation Models review, on-device AI audit, LanguageModelSession issues, @Generable checking, or Apple Intelligence integration review.

56

Quality

64%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./axiom-codex/skills/axiom-audit-foundation-models/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

77%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a highly actionable, well-sequenced auditor workflow with explicit validation and copy-paste fixes, but its length and inlined reference material (migration table, evaluations guidance) keep conciseness and progressive disclosure from the top band.

Suggestions

Move the OS27 error-migration table and the evaluations guidance (Patterns 11–13) into bundled reference files (e.g. references/os27-migration.md, references/evaluations.md) and link to them, leaving the body as the overview.

Deduplicate the grep patterns: have Phase 1 Steps collect the surface once and have Phase 2 Patterns reference those results instead of restating the regexes.

Convert the "axiom-ai (skills/...md)" pointers and the sibling-auditor list into proper markdown links and consolidate them into the Related section so they are clearly signaled once.

DimensionReasoningScore

Conciseness

The body is information-dense with no basic-concept padding, but it restates grep patterns between Phase 1 and Phase 2 and lists sibling auditors in both Phase 4 and the Related section, so it could be tightened beyond the "minor" threshold for a 4.

3 / 5

Actionability

Every detection pattern ships runnable grep regexes and copy-paste Swift fix examples (e.g. the availability guard, the streamResponse loop), plus a literal output template and a filled health-score table — fully executable guidance covering the common cases.

5 / 5

Workflow Clarity

A clear five-phase sequence with explicit validation checkpoints ("use Read to verify the surrounding context before reporting"), a phase gate ("Present this map in the output before proceeding"), a compound-severity feedback table, and a false-positive checklist for error recovery.

5 / 5

Progressive Disclosure

Sections are well-organized, but reference-grade material (the OS27 migration table, the evaluations deep-dive) is inlined in the ~440-line body rather than split into bundled reference files, and the external skill pointers are parenthetical text rather than clearly-signaled links.

3 / 5

Total

16

/

20

Passed

Description

51%Weight 40%Scale 1-5

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 has strong, distinctive trigger terms and a clear "Use when" clause, but it omits any statement of what the skill actually does, leaving the "what" half of completeness unanswered and capping specificity at the domain-naming level.

Suggestions

Lead with the skill's action: prepend "Detects Foundation Models (Apple Intelligence) anti-patterns and missing patterns in Swift codebases. Use when..." so the description answers both what and when.

Add the concrete failure categories the body already enumerates (availability/watchdog crashes, guardrail-refusal UX, prompt injection, structured-output breakage) to the description so the action is specific rather than implied.

Include natural variants like "Swift", "iOS 26 / macOS 26", and "Apple Intelligence audit" to broaden trigger-term coverage.

DimensionReasoningScore

Specificity

The description names concrete domains ("Foundation Models review", "on-device AI audit", "LanguageModelSession issues", "@Generable checking", "Apple Intelligence integration review") but states zero concrete actions the skill performs — it never says it detects/audits/fixes anything.

2 / 5

Completeness

The "when" is explicit ("Use when the user mentions..."), but the "what" is entirely absent — the description never states what the skill does, only when to invoke it, matching the "only when present without what" anchor.

2 / 5

Trigger Term Quality

Good keyword coverage including natural framework terms a developer would say ("Foundation Models", "on-device AI", "LanguageModelSession", "@Generable", "Apple Intelligence"), but missing some variants ("Swift", "iOS 26") and file extensions.

4 / 5

Distinctiveness Conflict Risk

Highly specific Apple framework identifiers ("LanguageModelSession", "@Generable", "Apple Intelligence") give this a clear niche with minimal overlap risk against other skills.

5 / 5

Total

13

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
CharlesWiltgen/Axiom
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.