audit-agent-overhead

Audit an agent setup (global ~/.claude AND project .claude AND plugin scope) for the 9 token-waste patterns that silently inflate cost and burn through usage limits. Outputs a per-pattern score against thresholds and proposes specific fixes — applied only with user approval. Heavyweight (~5KB SKILL.md + audit script); explicit invocation only. Triggers: /audit-agent-overhead, "audit my agent setup", "why am I hitting usage limits".

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

85%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.

Well-structured, concise, and clearly sequenced content with strong validation checkpoints. Its main defect is that the bundle it references (patterns.md, bin/audit.sh) is absent, which both undercuts actionability and breaks the progressive-disclosure navigation it otherwise sets up well.

Suggestions

Ship the referenced bundle: add patterns.md with the per-pattern fix recipes and bin/audit.sh so the central command and the offloaded detail actually resolve.

If the audit script is intentionally not bundled, state that explicitly and inline a minimal fallback (e.g. the grep/inspection commands the script would run) so the skill remains actionable without it.

Verify referenced paths resolve at load time; dangling links to patterns.md make the otherwise clean one-level-deep structure fail in practice.

Dimension	Reasoning	Score
Conciseness	Lean and efficient throughout — it assumes Claude's knowledge (no explaining what CLAUDE.md/MCPs are), uses a compact summary table, and the Anti-patterns section actively forbids verbose fix-docs; every section earns its tokens.	5 / 5
Actionability	Gives an executable command, exact paths, per-pattern thresholds, and a concrete per-fix workflow, but the central command targets `${CLAUDE_SKILL_DIR}/bin/audit.sh` which is not actually bundled, so the skill cannot be run as-shipped.	4 / 5
Workflow Clarity	A clear 5-step sequence (audit → score → propose fixes one at a time → hard rules → re-audit/report) with explicit validation checkpoints ('re-run the audit after each pattern to confirm the metric moved') appropriate to a destructive/batch operation.	5 / 5
Progressive Disclosure	The body is structured for one-level-deep disclosure (summary table pointing to patterns.md, detail offloaded), but the referenced bundle files — patterns.md and bin/audit.sh — do not exist, so the navigation resolves to nothing.	3 / 5
	Total	17 / 20 Passed

Description

92%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.

A strong, third-person description that clearly states what the skill does and when to invoke it, with concrete actions and natural trigger phrases. It is only marginally short of perfect on trigger-term coverage, where a few body-level synonyms are not surfaced in the description itself.

Dimension	Reasoning	Score
Specificity	Names the domain and multiple concrete actions — 'Audit an agent setup', 'Outputs a per-pattern score against thresholds', 'proposes specific fixes — applied only with user approval' — across three named scopes, giving comprehensive coverage.	5 / 5
Completeness	Explicitly answers both 'what' (audit 9 patterns, score vs thresholds, propose fixes) and 'when' via a dedicated 'Triggers:' clause with concrete phrases, matching the anchor for a clear what-and-when pair.	5 / 5
Trigger Term Quality	Includes natural user phrases ('audit my agent setup', 'why am I hitting usage limits') and the slash command, but a few common synonyms that appear in the body (e.g. 'trim my CLAUDE.md', 'Claude got dumber') are absent from the description.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (token-waste/overhead auditing of agent setups across three scopes) with distinct triggers and minimal overlap risk with other skills.	5 / 5
	Total	19 / 20 Passed

Validation

93%

Warnings & errors only

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

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
relative_links	Relative link issues: 2 missing	Warning

	Total	15 / 16 Passed

Repository: belchman/claude-skills
Path: plugins/audit-agent-overhead/skills/audit-agent-overhead/SKILL.md
Commit: 77a9e6b

Reviewed: 3 days ago

Table of Contents

Discovery Implementation Validation

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.