agent-architecture-audit

Full-stack diagnostic for agent and LLM applications. Audits the 12-layer agent stack for wrapper regression, memory pollution, tool discipline failures, hidden repair loops, and rendering corruption. Produces severity-ranked findings with code-first fixes. Essential for developers building agent applications, autonomous loops, or any LLM-powered feature.

Quality

70%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No known issues

Fix and improve this skill with Tessl

tessl review fix ./skills/agent-architecture-audit/SKILL.md

Quality

Content

72%

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

The content is highly actionable with concrete commands, models, and a report schema, and is well-organized as a self-contained skill. The main weaknesses are mild redundancy between diagnostic and failure-pattern sections and the absence of explicit validation checkpoints in the audit workflow.

Suggestions

Add an explicit checkpoint at the end of each audit phase (e.g. 'Before mapping: confirm you have file:line or log:row evidence for each suspected layer') to introduce feedback loops.

Consolidate 'Quick Diagnostic Questions' into 'Common Failure Patterns' to remove the token-redundant overlap between the two sections.

Optionally extract the full JSON report schema into a reference file and link to it, keeping the SKILL.md overview leaner.

Dimension	Reasoning	Score
Conciseness	The body is mostly lean and uses tables for density rather than explaining concepts Claude already knows, but the 'Quick Diagnostic Questions' section substantially overlaps the 'Common Failure Patterns' and 'Anti-Patterns' sections, adding tokens that restate the same ideas.	2 / 3
Actionability	Provides concrete, copy-paste-ready `rg` search commands, a concrete severity model, a default fix order, and a full executable JSON report schema — fully actionable guidance rather than abstract description.	3 / 3
Workflow Clarity	The four phases (Scope, Evidence Collection, Failure Mapping, Fix Strategy) are clearly sequenced, but there are no explicit validation checkpoints or feedback loops between phases confirming evidence is sufficient before mapping, or that mapping is complete before fix planning.	2 / 3
Progressive Disclosure	No bundle files exist and none are needed; the single SKILL.md is well-organized into clearly labeled sections (When to Activate, 12-Layer Stack, Failure Patterns, Audit Workflow, Severity Model, Report Schema), satisfying the guideline that skills with no need for external references can score 3 with well-organized sections.	3 / 3
	Total	10 / 12 Passed

Description

67%

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 is specific and well-differentiated, but is jargon-heavy and omits an explicit 'Use when...' trigger clause, which limits trigger-term quality and completeness. Adding natural-language triggers and a 'Use when' clause would raise the score.

Suggestions

Add an explicit 'Use when...' clause with natural triggers users would actually say, e.g. 'Use when the agent is getting worse, tools are flaky, or behavior degrades after adding wrapper/memory layers.'

Soften jargon in the description (e.g. reframe 'wrapper regression' and '12-layer agent stack') or pair each technical term with a plain-language symptom phrase.

Move a few user-facing symptom phrases currently in the body ('the agent is getting worse', 'tools are flaky') into the description to improve trigger-term coverage.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions: 'Audits the 12-layer agent stack for wrapper regression, memory pollution, tool discipline failures, hidden repair loops, and rendering corruption' and 'Produces severity-ranked findings with code-first fixes' — a comprehensive, specific action set rather than vague language.	3 / 3
Completeness	Clearly answers 'what does this do' but lacks an explicit 'Use when...' trigger clause in the description; per the judging guidelines, a missing explicit trigger guidance caps completeness at 2 even though intent is implied.	2 / 3
Trigger Term Quality	Includes some natural terms ('agent', 'LLM applications', 'tool calling', 'memory'), but leans heavily on internal jargon ('12-layer agent stack', 'wrapper regression') that users would not naturally say, and misses common user phrasings like 'the agent is getting worse' which appear only in the body, not the description.	2 / 3
Distinctiveness Conflict Risk	Occupies a clear niche (agent/LLM architecture auditing) with a named scope and explicit 'Do not use for' boundaries in the body, making it unlikely to trigger for the wrong skill.	3 / 3
	Total	10 / 12 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: affaan-m/ECC
Commit: 754b8dd

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.