{"slug": "agentguard-io-github-agentwares-agentguard-0-1-3", "title": "agentguard (io.github.agentwares/agentguard@0.1.3)", "summary": "AgentGuard v0.1.3, an MCP policy proxy from io.github.agentwares, was published to the official MCP registry on 2026-09-08, offering spend caps, approvals for destructive tools, a kill switch, dry-run, and audit log. The npm package @agentwares/agentguard passed a semantic screen by mcpindex.ai, which found no malicious instructions in its description, though the deterministic conformance probe has not yet been run.", "body_md": "[← Index](/leaderboard)\n\n# agentguard\n\nMCP policy proxy: spend caps, approvals for destructive tools, kill switch, dry-run, audit log.\n\nThe current version, v0.1.3, was published to the official MCP registry on 2026-09-08. It is distributed as @agentwares/agentguard on npm, filed under Monitoring & Logs by this index, and declares 2 environment variables. Removals and unreachable sources are measured across every server this index tracks, contract drift across the servers that answer in consecutive snapshots; [the index's current counts](/stats) put this record in context.\n\n## Using agentguard in Claude, Cursor, Gemini CLI, Cline, or Zed?\n\nMCP tool contracts can change remotely with no version bump. The mcpindex gate pins each contract and **HOLDs the call** when it drifts-before your agent acts. Zero credentials. This is not the package install for this server itself (use **Install this server** for that).\n\nRewrites your MCP host config so each server launches behind the gate. Inspect first: curl -fsSL https://mcpindex.ai/install.sh | less\n\n```\nuv tool install mcpindex-gate && mcpindex-config-wire\n```\n\n[method](/methodology)\n\nSemantic screen found no manipulation pattern in the description. Conformance probe not yet run.\n\n`mcpindex.integrity.description` passINFO\nevidence“No malicious instructions”via static_description\n\n- - Semantic screen only - the deterministic conformance probe has not run on this server\n- - Confidence is reported but not yet calibrated (v1)\n- - Screen reads the tool description, not the live behavior\n- - advisory\n- - registry description only no input schema\n- - screen model 8b\n- - The registry listing (version, packages, links) changed after this screen ran - the description we assessed did not\n\nSemantic screen: an LLM judge reads the tool description for hidden instructions (status PARTIAL). A pass means the description is not lying, not that the tool is safe: a high-capability tool with an honest description still warrants caution. The deterministic conformance probe has not been run on this server yet, so the screen here is semantic-only. Posture: advisory. Confidences are reported but not yet calibrated (calibrated=false at v1). Full verdict history is not shown on this page.\n\nOwn this server? [Screen its description →](/screen)\n\n## That verdict was true at screening time (snapshot 2026-09-08).\n\nContracts can change after screening, with no version bump. The gate pins agentguard’s tool contracts on first sight and holds any silent change before your agent acts - the check that keeps being true on Tuesday.\n\n[See your first HOLD in 2 minutes →](/guides/install-the-gate-first-hold)\n\nRelated: [how to trust an MCP server](/guides/how-to-trust-an-mcp-server) · [screen before install](/guides/screen-mcp-server-before-install) · [silent contract drift](/guides/mcp-silent-contract-drift)\n\nPeople vet a server before wiring it in. One line on your project site or docs answers that with an independent record: screening verdict, registry provenance, contract drift history. It stays current as new screens and drift events land.\n\n```\n[Independent trust record for agentguard](https://mcpindex.ai/server/io-github-agentwares-agentguard) - screening verdict, registry provenance, and contract drift monitoring.\nIndependent trust record for agentguard - screening verdict, registry provenance, and contract drift monitoring.\n```\n\nA live verdict badge for your README or listing. It reflects the current screen, links back here, and updates when the verdict does.\n\n```\n[](https://mcpindex.ai/server/io-github-agentwares-agentguard)\n![\"mcpindex](\"https://mcpindex.ai/api/v1/badge/io-github-agentwares-agentguard\")
\n```\n\n`AGENTGUARD_CONFIG`\nPath to agentguard.yaml (default ./agentguard.yaml)\n\n`AGENTGUARD_KILL`\nSet to 1 to halt every tool call (kill switch)\n\n[methodology](/methodology)\n\n[Labbyai.dinglebear/labby](/server/ai-dinglebear-labby)\n\nRust MCP gateway with Code Mode, authentication, setup, logs, CLI, HTTP API, and operator web UI.\n\n[Maniloapp.manilo/manilo](/server/app-manilo-manilo)\n\nLog, query, and edit expenses, budgets, and accounts in Manilo from any MCP-compatible AI assistant.\n\n[Waqi — AI Privacy Layerio.github.ajprolific/waqi](/server/io-github-ajprolific-waqi)\n\nRedacts PII from your business tools before the AI sees them — with a provable audit log.", "url": "https://wpnews.pro/news/agentguard-io-github-agentwares-agentguard-0-1-3", "canonical_source": "https://mcpindex.ai/server/io-github-agentwares-agentguard", "published_at": "2026-09-08 00:17:02+00:00", "updated_at": "2026-09-08 01:00:43.416751+00:00", "lang": "en", "topics": ["ai-tools", "ai-safety", "ai-infrastructure"], "entities": ["AgentGuard", "io.github.agentwares", "@agentwares/agentguard", "mcpindex.ai"], "alternates": {"html": "https://wpnews.pro/news/agentguard-io-github-agentwares-agentguard-0-1-3", "markdown": "https://wpnews.pro/news/agentguard-io-github-agentwares-agentguard-0-1-3.md", "text": "https://wpnews.pro/news/agentguard-io-github-agentwares-agentguard-0-1-3.txt", "jsonld": "https://wpnews.pro/news/agentguard-io-github-agentwares-agentguard-0-1-3.jsonld"}}