AgentStack
SKILL verified MIT Self-run

Marketplace Audit

skill-acaprino-claude-code-daodan-marketplace-audit · by acaprino

>

No reviews yet
0 installs
16 views
0.0% view→install

Install

$ agentstack add skill-acaprino-claude-code-daodan-marketplace-audit

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Marketplace Audit? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Marketplace Audit

Run a comprehensive structural validation of any Claude Code plugin marketplace. Works against any project that follows the standard .claude-plugin/marketplace.json + plugins// layout.

Audit steps

Step 1: Run the validation script

Execute the audit script to get a machine-readable report:

# Validate only
python "${CLAUDE_PLUGIN_ROOT}/skills/marketplace-audit/scripts/audit_marketplace.py"

# Validate and auto-fix color issues (invalid, missing, disharmonious)
python "${CLAUDE_PLUGIN_ROOT}/skills/marketplace-audit/scripts/audit_marketplace.py" --fix

The script resolves the target project root by walking up from the script location, or respects a --project-root flag if invoking it from a different marketplace than where the plugin is installed.

Step 2: Review findings

The script checks:

  1. File existence -- every path in marketplace.json agents / skills / commands arrays resolves to a real file or directory
  2. Orphaned files -- agent .md files, skill directories, or command .md files on disk not registered in any plugin
  3. Frontmatter validation
  • Agents: must have name, description, model, color
  • Skills: must have name, description
  • Commands: must have description
  1. Color consistency and harmony
  • All agents within a plugin should use the same color
  • Warn when a single color is overused across too many plugins (threshold configurable)
  • Report color distribution across all plugins
  • Valid colors: red, blue, green, yellow, purple, orange, pink, cyan
  • Use --fix to auto-correct invalid or missing colors
  1. Naming conventions
  • All names are kebab-case
  • Agent filename matches frontmatter name field
  • Plugin directory name matches marketplace.json name field
  • Skill directory name matches frontmatter name field
  • Workflow command output directories match command filename (e.g., feature-e2e.md typically writes to .feature-e2e/)
  • No naming collisions between commands in different plugins
  • No em dash characters anywhere (use hyphen - or double hyphen --)
  1. Cross-reference consistency (project-specific, best-effort)
  • If a git remote is configured, suggest that the marketplace name align with the repo name (warning only)
  • If a CLAUDE.md is present at the project root, suggest that its project header match the marketplace name (warning only)
  • These are advisory -- different marketplace maintainers have different conventions
  1. Marketplace.json schema
  • Every plugin has: name, source, description, version, author, license, keywords, category, strict
  • No duplicate plugin names
  • Duplicate keywords across plugins (warning only -- may be intentional)
  1. Version sanity
  • All versions are valid semver (MAJOR.MINOR.PATCH)
  • metadata.version is present at the root

Step 3: Fix issues

Address findings by severity:

  • CRITICAL: Missing referenced files, broken paths, missing required frontmatter fields, duplicate plugin names
  • WARNING: Orphaned files, naming mismatches, overlapping keywords, color inconsistencies, git/CLAUDE.md alignment
  • INFO: Suggestions for improvement, consolidation opportunities

Step 4: Evaluate color harmony

After the script passes, review the color distribution and evaluate semantic harmony:

  1. Read each plugin's description and category from marketplace.json
  2. Consider domain: similar domains should have visually related colors; distinct domains should contrast
  3. Guiding principles:
  • Warm colors (red, orange, yellow, pink) for creative / outward-facing plugins
  • Cool colors (blue, cyan, purple) for analytical / development plugins
  • Neutral (green) for tooling / infrastructure
  1. If colors feel disharmonious, propose a new assignment with reasoning and apply after user confirmation

Step 5: Re-validate

Run the script again after fixes to confirm a clean audit.

Notes for marketplace maintainers

  • This skill is marketplace-agnostic. It assumes the standard Claude Code plugin layout (.claude-plugin/marketplace.json + plugins//) but makes no assumption about which plugins, authors, or upstream sources are specific to your marketplace.
  • Plugin categories, default author, upstream sources, and any project-specific conventions should be documented in your CLAUDE.md at the project root. The script and this skill respect those conventions but do not enforce a particular taxonomy.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.