Best for
- Reviewing what tools and commands are pre-allowed
- Cleaning up accumulated ad-hoc permission entries
- After adding a new MCP server and deciding which operations to pre-allow
jackchuka/skills/claude-permissions-audit/SKILL.md
Review and reorganize Claude Code permission settings across all config files (global settings.json, project settings.local.json, dotfiles copies). Identifies redundancy, misplaced permissions, and lack of read/write organization. Produces a clean layout where global settings are the source of truth and project-local files only contain project-specific overrides. Use this skill whenever the user mentions reviewing permissions, cleaning up settings, auditing allowed tools, reorganizing settings.j
Decision brief
Review and reorganize Claude Code permission settings so global settings are the source of truth, project-local files stay minimal, and read vs write operations have clear separation.
Compatibility matrix
| Platform | Status | Evidence | What to check |
|---|---|---|---|
| Codex | Not declared | No explicit evidence | Portability before use |
| Claude Code | Declared | Source record | Install path and trigger |
| Cursor | Not declared | No explicit evidence | Portability before use |
| Gemini CLI | Not declared | No explicit evidence | Portability before use |
Installation
The source command is displayed only when detected. A safe inspection prompt is always available so your agent can explain every action before execution.
npx skills add https://github.com/jackchuka/skills --skill "claude-permissions-audit"Inspect the Agent Skill "claude-permissions-audit" from https://github.com/jackchuka/skills/blob/7b0b33f68b8f11522e43622e5cb3bacd802999d2/claude-permissions-audit/SKILL.md at commit 7b0b33f68b8f11522e43622e5cb3bacd802999d2. List every install step, command, network request, credential, file read/write, external action, and rollback step. Explain whether it fits my task. Do not install or execute anything until I approve.
Workflow
Scan for every settings file that contributes to the effective permission set.
For each permission entry across all files, assess:
Design the new layout following these principles:
Once the user confirms the plan:
Present a before/after comparison:
Permission review
The documentation asks the agent to read local files, directories, or repositories.
Scan for every settings file that contributes to the effective permission set.The documentation asks the agent to read local files, directories, or repositories.
Read all of them in parallel. If a file doesn't exist, note it and move on.The documentation asks the agent to create, modify, or delete local files.
Write the updated global settings fileEvidence record
| Signal | Value | Evidence type | Meaning |
|---|---|---|---|
| Quality score | 88/100 | Computed | Documentation, specificity, maintenance, and trust rules |
| Repository stars | 15 | Source | Repository attention, not individual Skill quality |
| Compatibility | 1 platforms | Source | Declared in the catalog source record |
| Usage guide | automated source guide | Editorial | Generated or reviewed according to the visible evidence level |
Pinned source
Review and reorganize Claude Code permission settings so global settings are the source of truth, project-local files stay minimal, and read vs write operations have clear separation.
Scan for every settings file that contributes to the effective permission set.
Locations to check:
~/.claude/settings.json — global settings (source of truth)<project>/.claude/settings.json — project shared settings (committed)<project>/.claude/settings.local.json — project local settings (gitignored)~/.claude/.claude/settings.local.json — home directory project localhome/.claude/settings.json in a dotfiles repo)Read all of them in parallel. If a file doesn't exist, note it and move on.
For each permission entry across all files, assess:
Bash(gws calendar events:*) is redundant when Bash(gws:*) exists globallyBash(python3:*) in project local when it's useful everywheregh pr list, gh pr view individually, but project local has gh:*Present findings as a table with issues and proposed resolution for each.
Design the new layout following these principles:
Organize permissions into labeled sections by grouping related entries together. Do NOT use JSON comments (// or /* */) — settings.json must be valid JSON. Instead, rely on the ordering of entries to convey grouping. Within the allow array, cluster related permissions together in this order:
Section order (group entries in this sequence, no separators needed):
filesystem reads — cat, tree, grep, wc, file reads
<lang> inspection — go doc, cargo check, linters (no side effects)
GitHub CLI reads — gh pr list/view/diff, gh issue list (no mutations)
external tool reads — calendar, slack CLI, other read CLIs
web reads — WebSearch, WebFetch with domain allowlist
MCP <service> reads — list, get, describe, search operations
build / test / install — go build, cargo test, pnpm install (local side effects)
skills — Skill(superpowers:*), etc.
Only keep entries that are genuinely project-specific:
Bash(gh:*) for a repo that creates PRs frequently)Before applying, ask the user about edge cases. Common questions:
gh:* (all subcommands) go global, or keep read-only globally with broader access per-project?Adapt questions to what's actually ambiguous — don't ask about things that are obvious from context.
Once the user confirms the plan:
Present a before/after comparison:
| File | Before | After |
|---|---|---|
| Global | N flat entries | M entries in K sections |
| Project local | X entries | Y entries (project-specific only) |
List what moved where, and what still requires prompting (write operations not pre-allowed).
Alternatives
jackchuka/skills
Audit skill SKILL.md files for compliance with the agentskills.io specification and house conventions. Checks frontmatter fields (name, description, compatibility, metadata, argument-hint), metadata sub-fields (author, scope, layer, confirms), and layer/suffix consistency. Use when adding new skills, reviewing skill quality, or ensuring all skills follow the spec. Triggers: "audit skills", "check skill spec", "skill compliance", "are my skills up to spec", "/claude-skill-spec-audit".
affaan-m/ECC
Patterns and architectures for autonomous Claude Code loops — from simple sequential pipelines to RFC-driven multi-agent DAG systems.
affaan-m/ECC
GAN-inspired Generator-Evaluator agent harness for building high-quality applications autonomously. Based on Anthropic's March 2026 harness design paper.
affaan-m/ECC
Write-time code quality enforcement using Plankton — auto-formatting, linting, and Claude-powered fixes on every file edit via hooks.