{
  "markdown": "# agentguard\n\nA policy proxy between an AI agent and the tools it can reach: hard spend limits,\napproval before anything destructive, a kill switch, per-agent scoped credentials,\ndry-run writes with a mutation diff, a semantic loop breaker, and a hash-chained\naudit log you can verify.\n\n```bash\nnpx @agentwares/agentguard init\n```\n\nNo LLM calls anywhere in it. No phone-home. No account. MIT.\n\n- `packages/agentguard-core` — the policy engine (Web-standard, runs anywhere)\n- `apps/agentguard-cli` — `agentguard`, the proxy and its commands\n- `packages/agentguard-sdk` — the same engine for non-MCP tool calls\n- `assets/permission-diff-action` — a GitHub Action that comments when agent permissions widen\n- `packages/mcp-kit`, `packages/notify`, `packages/x402` — generic plumbing the CLI bundles\n\nHosted incident dashboard, audit retention and alerting: https://agentwares-agentguard.vercel.app\n\nThis repository is exported from a private monorepo, which is why it has no long history.\nThe exported source is complete and is exactly what is published to npm.\n",
  "bytes": 1059,
  "sha": "89b6c00e2ed7378cad7dd2c78a4cf669ca09192147c903c8c783131f81971a80",
  "repo_slug": "agentwares/agentguard",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_agentwares_agentguard_5388d9f2/readme"
}