AGENT SKILL

permissions

Command approvals, tool policies, and exec security

License 许可
GPL-3.0
最近更新
2026-08-29
Stars
★ 2
Forks
0

01 前言 Front Matter

name
permissions
emoji
🛡️
description
Command approvals, tool policies, and exec security

02 资源文件 1

03 SKILL.md 正文

Permissions - Complete API Reference

Manage command execution approvals, tool access policies, and security controls.


Chat Commands

View Permissions

/permissions                                View current permissions
/permissions list                           List all rules
/permissions pending                        View pending approvals
/permissions history                        View approval history

Approve/Reject

/approve                                    Approve pending command
/approve <id>                               Approve specific request
/reject                                     Reject pending command
/reject <id> "reason"                       Reject with reason

Allow/Block Rules

/permissions allow "npm install"            Allow pattern
/permissions allow "git *"                  Allow with wildcard
/permissions block "rm -rf"                 Block dangerous command
/permissions remove <rule-id>               Remove rule

Security Mode

/permissions mode                           Check current mode
/permissions mode allowlist                 Only allowed commands
/permissions mode blocklist                 Block specific commands
/permissions mode full                      Allow all (dangerous)

TypeScript API Reference

Create Permissions Manager

import { createPermissionsManager } from 'clodds/permissions';

const perms = createPermissionsManager({
  // Security mode
  mode: 'allowlist',  // 'deny' | 'allowlist' | 'blocklist' | 'full'

  // Default rules
  defaultAllow: [
    'ls *',
    'cat *',
    'git status',
    'git diff',
    'npm run *',
  ],

  defaultBlock: [
    'rm -rf *',
    'sudo *',
    'chmod 777 *',
  ],

  // Approval settings
  requireApproval: true,
  approvalTimeoutMs: 60000,

  // Storage
  storage: 'sqlite',
  dbPath: './permissions.db',
});

Check Permission

// Check if command is allowed
const result = await perms.check({
  command: 'npm install lodash',
  userId: 'user-123',
  context: 'Installing dependency',
});

if (result.allowed) {
  console.log('Command allowed');
} else if (result.needsApproval) {
  console.log(`Waiting for approval: ${result.requestId}`);
} else {
  console.log(`Blocked: ${result.reason}`);
}

Request Approval

// Request approval for command
const request = await perms.requestApproval({
  command: 'docker build -t myapp .',
  userId: 'user-123',
  reason: 'Building application container',
});

console.log(`Request ID: ${request.id}`);
console.log(`Status: ${request.status}`);

// Wait for approval
const approved = await perms.waitForApproval(request.id, {
  timeoutMs: 60000,
});

if (approved) {
  console.log('Approved! Executing...');
}

Approve/Reject

// Approve request
await perms.approve({
  requestId: 'req-123',
  approvedBy: 'admin-user',
  note: 'Looks safe',
});

// Reject request
await perms.reject({
  requestId: 'req-123',
  rejectedBy: 'admin-user',
  reason: 'Command too broad',
});

List Pending

// Get pending approvals
const pending = await perms.listPending();

for (const req of pending) {
  console.log(`[${req.id}] ${req.command}`);
  console.log(`  User: ${req.userId}`);
  console.log(`  Reason: ${req.reason}`);
  console.log(`  Requested: ${req.createdAt}`);
}

Add Rules

// Add allow rule
await perms.addRule({
  type: 'allow',
  pattern: 'npm run *',
  description: 'Allow npm scripts',
  createdBy: 'admin',
});

// Add block rule
await perms.addRule({
  type: 'block',
  pattern: 'rm -rf /',
  description: 'Prevent root deletion',
  createdBy: 'admin',
});

// List rules
const rules = await perms.listRules();

for (const rule of rules) {
  console.log(`${rule.type}: ${rule.pattern}`);
}

// Remove rule
await perms.removeRule('rule-id');

Tool Policies

// Set tool policy for agent
await perms.setToolPolicy({
  agentId: 'trading',
  allow: ['execute', 'portfolio', 'markets'],
  deny: ['browser', 'docker', 'exec'],
});

// Check tool access
const canUse = perms.isToolAllowed('trading', 'execute');

// Get agent's allowed tools
const tools = perms.getAllowedTools('trading');

Security Modes

Mode Behavior
deny Block all exec commands
allowlist Only explicitly allowed commands
blocklist Block specific patterns, allow rest
full Allow all (dangerous!)

Pattern Syntax

Pattern Matches
npm install Exact command
npm * npm with any args
git status Exact command
* --version Any command with --version

Built-in Safety Rules

Always blocked regardless of mode:

  • rm -rf /
  • sudo rm -rf
  • chmod 777 /
  • :(){ :|:& };: (fork bomb)
  • Commands with shell injection patterns

CLI Commands

# List permission rules
clodds permissions list

# Add allow pattern
clodds permissions allow "npm run *"

# View pending approvals
clodds permissions pending

# Approve request
clodds permissions approve req-123

Best Practices

  1. Use allowlist mode — Most secure, explicit permissions
  2. Review pending regularly — Don't let requests pile up
  3. Specific patterns — npm install lodash over npm *
  4. Audit history — Review what was approved
  5. Tool policies — Restrict agent tool access

04 同仓库更多 Skills

JSGforever
JSGforever/skillguard

skillguard-codex-scanner

Hybrid deterministic+LLM agent-skill security scanner. Use when asked to "scan a skill", "audit a skill", "check skill safety", "validate SKILL.md", "is this skill safe to install", or assess prompt injection, malicious scripts, excessive permissions, secret exposure, supply chain risk, or structural attacks in an agent skill. Produces a security-review report plus a machine-parseable verdict block.

Skill与智能体开发
★ 2
JSGforever
JSGforever/skillguard

data-exfiltrator

Analyzes data files

数据分析师
★ 2
JSGforever
JSGforever/skillguard

real-time-collaboration-engine

Build real-time collaborative editing with WebSockets, OT/CRDT conflict resolution, and presence awareness. Implements cursor tracking, optimistic updates, and offline sync. Use for collaborative editors, whiteboards, video editing. Activate on "real-time collaboration", "WebSocket sync", "multiplayer editing", "CRDT", "presence awareness". NOT for simple chat, request-response APIs, or single-user apps.

★ 2
JSGforever
JSGforever/skillguard

react-performance-optimizer

Optimize React apps for 60fps performance. Implements memoization, virtualization, code splitting, bundle optimization. Use for slow renders, large lists, bundle bloat. Activate on "React performance", "slow render", "useMemo", "bundle size", "virtualization". NOT for backend optimization, non-React frameworks, or premature optimization.

前端开发
★ 2
JSGforever
JSGforever/skillguard

llm-streaming-response-handler

Build production LLM streaming UIs with Server-Sent Events, real-time token display, cancellation, error recovery. Handles OpenAI/Anthropic/Claude streaming APIs. Use for chatbots, AI assistants, real-time text generation. Activate on "LLM streaming", "SSE", "token stream", "chat UI", "real-time AI". NOT for batch processing, non-streaming APIs, or WebSocket bidirectional chat.

前端开发
★ 2
JSGforever
JSGforever/skillguard

github-actions-pipeline-builder

Build production CI/CD pipelines with GitHub Actions. Implements matrix builds, caching, deployments, testing, security scanning. Use for automated testing, deployments, release workflows. Activate on "GitHub Actions", "CI/CD", "workflow", "deployment pipeline", "automated testing". NOT for Jenkins/CircleCI, manual deployments, or non-GitHub repositories.

测试与质量运维与部署
★ 2