permission-tuner

$npx mdskill add rohitg00/pro-workflow/permission-tuner

Reduce permission prompts by identifying and applying safe alwaysAllow rules

  • Identify frequently-approved tools for automatic inclusion in permissions.
  • Deploys a list of safe `alwaysAllow` rules to automatically grant access.
  • Decides which tools are low-risk candidates based on their functionality.
  • Delivers the final set of `alwaysAllow` rules directly to the user's settings.

SKILL.md

.github/skills/permission-tunerView on GitHub ↗
---
name: permission-tuner
description: Analyze permission denial patterns and generate optimized alwaysAllow and alwaysDeny rules. Use when permission prompts are slowing you down or after sessions with many denials.
user-invocable: true
---

# Permission Tuner

Reduce permission prompt fatigue by analyzing denial patterns and suggesting targeted rules.

## Trigger

Use when:
- Permission prompts interrupt flow repeatedly
- Starting a new project and want to configure permissions
- After a session with many manual approvals

## Workflow

1. Scan recent session data for permission patterns
2. Identify frequently-approved tools and patterns
3. Generate safe `alwaysAllow` rules
4. Present rules for approval before applying

## Analysis

### Step 1: Gather Permission Data

Check current permission rules:
```bash
cat .claude/settings.json 2>/dev/null | grep -A 20 "permissions"
cat ~/.claude/settings.json 2>/dev/null | grep -A 20 "permissions"
```

### Step 2: Identify Safe Patterns

**Allow-list candidates** (low risk):
- `Read` — all file reads (read-only, no side effects)
- `Glob` — file pattern matching (read-only)
- `Grep` — content search (read-only)
- `Bash(git status)` — read-only git commands
- `Bash(git diff*)` — read-only git commands
- `Bash(git log*)` — read-only git commands
- `Bash(npm test*)` — test execution
- `Bash(npm run lint*)` — linting
- `Bash(npm run typecheck*)` — type checking

**Ask candidates** (medium risk — prompt user every time):
- `Edit` — file modifications
- `Write` — new file creation
- `Bash(git add*)` — staging changes
- `Bash(git commit*)` — creating commits
- `Bash(npm install*)` — dependency changes

**Deny-list candidates** (high risk):
- `Bash(git push*)` — affects remote
- `Bash(git reset --hard*)` — destructive
- `Bash(rm -rf*)` — destructive
- `Bash(curl*POST*)` — external API calls
- Any command with `--force` or `--no-verify`

### Step 3: Generate Rules

```json
{
  "permissions": {
    "allow": [
      "Read",
      "Glob",
      "Grep",
      "Bash(git status)",
      "Bash(git diff*)",
      "Bash(git log*)",
      "Bash(npm test*)",
      "Bash(npm run lint*)",
      "Bash(npm run typecheck*)"
    ],
    "deny": [
      "Bash(rm -rf *)",
      "Bash(git push --force*)",
      "Bash(git reset --hard*)"
    ]
  }
}
```

## Output

```text
PERMISSION TUNER REPORT

Current rules: [X] allow, [Y] deny, [Z] ask

Recommendations:
  Auto-approve (safe, read-only):
    + Read, Glob, Grep
    + Bash(git status), Bash(git diff*), Bash(git log*)

  Auto-approve (medium risk, frequently used):
    + Edit (approved X times this session)
    + Bash(npm test*) (approved X times)

  Keep asking:
    ~ Bash(git commit*) — verify commit messages
    ~ Write — verify new file creation

  Auto-deny (dangerous):
    - Bash(rm -rf *)
    - Bash(git push --force*)

Estimated prompts saved per session: ~[N]
```

## Rules

- Destructive operations must stay in the deny list
- Always present rules for user approval before applying
- Group rules by risk level (safe/medium/dangerous)
- Include estimated prompt savings

More from rohitg00/pro-workflow

SkillDescription
agent-teamsCoordinate multiple Claude Code sessions as a team — lead + teammates with shared task lists, mailbox messaging, and file-lock claiming. Patterns for team sizing, task decomposition, and when to use teams vs sub-agents vs worktrees.
auto-setupAuto-configure quality gates, hooks, and settings for a new project. Detects project type and sets up appropriate tooling. Use when onboarding a new codebase.
batch-orchestrationDecompose large-scale changes into independent units and spawn parallel agents in isolated worktrees. Use for migrations, refactors, codemods, and any change touching 10+ files with the same pattern.
bug-captureCapture a user-reported defect as a durable GitHub issue written in the project's own domain language. Explores the codebase in parallel for context but never leaks file paths or line numbers into the issue. Use when the user reports a bug conversationally, runs a QA pass, or says "file an issue", "log this as a bug", "capture this".
compact-guardSmart context compaction with state preservation. Saves critical files, task progress, and working state before compaction, restores after. Use before manual compact or when auto-compact triggers.
context-optimizerOptimize token usage and context management. Use when sessions feel slow, context is degraded, or you're running out of budget.
cost-trackerTrack session costs, set budget alerts, and optimize token spend. Use to check costs mid-session or set spending limits.
design-engineeringApply interface craft when building or reviewing UI - motion, easing, timing, springs, component feel, and visual foundations. Use when building a component, animation, transition, hover or press state, modal, drawer, toast, or when polishing an interface so it feels right. Says "make this feel better", "add an animation", "polish the UI", "review this component".
deslopRemove AI-generated code slop, unnecessary comments, and over-engineering from the current branch diff. Cleans up boilerplate, simplifies abstractions, strips defensive code, and in skill-file mode lints SKILL.md files for quality. Use when cleaning up code, simplifying, removing boilerplate, before committing, or when reviewing a skill before promoting it.
domain-modelingBuild the project's shared language and bounded contexts before writing code, so names stay consistent and the agent stops paraphrasing domain concepts. Produces a CONTEXT.md glossary and decision records. Use at the start of a project or feature, or when the codebase and the people describing it speak different languages.