Metacognitive AI agent oversight: adaptive CPI interrupts for alignment, reflection and safety
If you want an AI assistant working directly with Vibe Check, the vibe check mcp server is the bridge. Metacognitive AI agent oversight: adaptive CPI interrupts for alignment, reflection and safety.
KISS overzealous agents goodbye. Plug & play agent oversight tool.
Vibe Check MCP keeps agents on the minimal viable path and escalates complexity only when evidence demands it. Vibe Check MCP is a lightweight server implementing Anthropic's Model Context Protocol. It acts as an AI meta-mentor for your agents, interrupting pattern inertia with Chain-Pattern Interrupts (CPI) to prevent Reasoning Lock-In (RLI). Think of it as a rubber-duck debugger for LLMs – a quick sanity check before your agent goes down the wrong path.
Once connected, the assistant can call these 14 tools directly:
Security — ** npm audit is clean — 10 advisories resolved across axios, the MCP SDK's Hono stack, form-data, fast-uri, postcss and the test toolchainDependencies — ** MCP SDK 1.29, axios 1.18, OpenAI SDK 6.x, vitest 4.x; the unused body-parser direct dependency was droppedFeature — DescriptionProvider — Default modelgemini — gemini-3.6-flashanthropic — claude-sonnet-5openai — gpt-5.6-terraopenrouter — (none — required)Variable — DefaultCORS_ORIGIN — loopback origins onlyMCP_ALLOWED_HOSTS — localhost, 127.0.0.1, ::1MCP_MAX_BODY_SIZE — 100kbRecognition — The Recognition tool exposed by this serverConfiguration — See docs/TESTING.md for instructions on how to run testsThe server is distributed via npm as @pv-bhat/vibe-check-mcp, so most clients can run it without a manual build step. Add it to your MCP client's configuration and restart the client to pick it up — the copy-paste configs for Claude Desktop, Claude Code and Cursor are on this page.
Before the server will start you need to supply 6 environment variables: GEMINI_API_KEY, OPENAI_API_KEY, OPENROUTER_API_KEY, ANTHROPIC_API_KEY, ANTHROPIC_AUTH_TOKEN, ANTHROPIC_BASE_URL. Keep credentials in your client's env block or a secrets manager rather than committing them.
Developer-tool servers are usually the first ones people connect, because they turn "help me with this code" into an assistant that can actually read the repo and act on it. Vibe Check sits in that group, and the shape of its toolset — Security, Dependencies, Feature among others — tells you what it is really for. Worth comparing against the other developer tools servers in this directory before you commit to one, since several overlap in scope but differ sharply in setup cost and permissions.
| Tool | What it does |
|---|---|
| Security | ** npm audit is clean — 10 advisories resolved across axios, the MCP SDK's Hono stack, form-data, fast-uri, postcss and the test toolchain |
| Dependencies | ** MCP SDK 1.29, axios 1.18, OpenAI SDK 6.x, vitest 4.x; the unused body-parser direct dependency was dropped |
| Feature | Description |
| Provider | Default model |
| gemini | gemini-3.6-flash |
| anthropic | claude-sonnet-5 |
| openai | gpt-5.6-terra |
| openrouter | *(none — required)* |
| Variable | Default |
| CORS_ORIGIN | loopback origins only |
| MCP_ALLOWED_HOSTS | localhost, 127.0.0.1, ::1 |
| MCP_MAX_BODY_SIZE | 100kb |
| Recognition | The Recognition tool exposed by this server. |
| Configuration | See [docs/TESTING.md]() for instructions on how to run tests. |
{
"mcpServers": {
"vibe-check": {
"command": "npx",
"args": ["-y", "@pv-bhat/vibe-check-mcp"],
"env": {
"GEMINI_API_KEY": "your-value",
"OPENAI_API_KEY": "your-value",
"OPENROUTER_API_KEY": "your-value",
"ANTHROPIC_API_KEY": "your-value",
"ANTHROPIC_AUTH_TOKEN": "your-value",
"ANTHROPIC_BASE_URL": "your-value"
}
}
}
}Add to claude_desktop_config.json, then restart Claude Desktop.
| Variable | Description | Required |
|---|---|---|
| GEMINI_API_KEY | Credential the server authenticates with. | Yes |
| OPENAI_API_KEY | Credential the server authenticates with. | Yes |
| OPENROUTER_API_KEY | Credential the server authenticates with. | Yes |
| ANTHROPIC_API_KEY | Credential the server authenticates with. | Yes |
| ANTHROPIC_AUTH_TOKEN | Credential the server authenticates with. | Yes |
| ANTHROPIC_BASE_URL | Endpoint or connection string the server talks to. | Yes |
Kill hallucinated APIs — version-accurate, up-to-date library documentation injected straight into context.
Microsoft's official browser automation server — drive a real browser through the accessibility tree, no screenshots needed.
GitHub's official server — repos, issues, pull requests, Actions and code security, straight from your assistant.
Issue tracking at the speed of conversation — Linear's official hosted server with OAuth and zero install.
Local repository surgery — status, diffs, commits, branches and history for any repo on disk.
Timezone sanity for AI — current time anywhere and correct conversions, without the model doing date math.