High-risk tools in 0nmcp

Chat with your local Llama model. Zero cost, fully private. Supports multi-turn conversation, system prompts, and .brain file loading. Example: ai_chat({ message:

Create text embeddings locally using Ollama. Free, private. Useful for: semantic search, similarity matching, RAG pipelines. Example: ai_embed({ text:

One-shot text generation with local Llama. No conversation context. Good for: summarization, code generation, analysis, content creation. Example: ai_generate({ prompt:

Make a direct API call to any connected service. For advanced use when you need fine-grained control beyond the execute tool.

Build a .0n application bundle — a portable encrypted file containing endpoints, workflows, operations, automations, and connections. Deploy anywhere with: 0nmcp app run <file> ...

Open a .0n application bundle file. Decrypts and extracts the application for local use. Example: app_open({ bundle:

Compile a trained brain into a portable .brain file. The .brain file can be imported into any app on any LLM. Also generates the system prompt version for direct use. Example:...

Import a .brain file and extract its system prompt for use in any LLM. Can import from file path or raw JSON. Example: brain_import({ path:

Run scenario-based training on a brain. Executes each scenario against the brain

Have AI providers critique each other

Full multi-AI problem-solving pipeline: 1. Ask all AIs the question (parallel) 2. Have them critique each other 3. Synthesize the best answer from all responses This is the mos...

Build and deploy a complete AI Workflow (Agent Studio agent) from a natural language description. Creates the agent, configures knowledge base connections, adds MCP server nodes...

Deploy a full snapshot — pipeline, tags, custom values, and workflow definitions — in a single operation.

Execute an AI Workflow (Agent Studio agent). Send a message and get a response. Maintains conversation context via executionId.

Execute any task using connected services. The AI orchestrator automatically: 1. Parses your intent from natural language 2. Finds the best services to use 3. Creates an executi...

Fire one action immediately — no flow, no enrollment, no schedule. Useful for one-shot sends (

POST the rendered HTML to Vercel as a single-file static deployment. Returns the live URL. Requires VERCEL_TOKEN in connections or passed as arg.

Build a plugin from a service key or custom spec. If building from catalog, returns the plugin

Execute a plugin endpoint with automatic .0n field resolution. Accepts canonical .0n fields (email.0n, fullname.0n, etc.) and auto-translates to the service

Mirror an rc_sequence into user_workflows so the customer can edit / run / schedule it from the onork-app /dashboard/automations surface. Produces a node/edge graph compatible w...

Execute a pre-defined .0n workflow file. Workflows are deterministic, step-by-step automations stored in ~/.0n/workflows/. Unlike the

Execute a plan that a human has approved. Runs only the legs carried in the signed plan. A plan can be run once — replay is blocked in the database, not in code. Every leg writ...

Manage the 0nAI training feed — continuous data ingestion from verified public sources. Fetches from ${FEED_SOURCES.length} sources: Hacker News, arXiv, Dev.to, GitHub, npm, Coi...

Unwrap a 0nVault container using an escrow party

Decrypt a service