# ShoulderTap

## Docs

- [Audit](https://docs.shouldertap.dev/api/audit.md): Retrieve the full event trail for a single request over the ShoulderTap HTTP API.
- [Authentication](https://docs.shouldertap.dev/api/authentication.md): The ShoulderTap HTTP API uses a single static bearer token, read from an environment variable. Learn how to configure and send it.
- [Consumers](https://docs.shouldertap.dev/api/consumers.md): Register and unregister webhook consumers over the ShoulderTap HTTP API.
- [Contracts](https://docs.shouldertap.dev/api/contracts.md): The wire shapes ShoulderTap clients agree on — ContextRequest, ContextProposal, and ConsumerRegistration — with every field.
- [Status codes and reason codes](https://docs.shouldertap.dev/api/errors.md): HTTP status codes the ShoulderTap API returns, and the reason codes carried on a failed request.
- [Experts](https://docs.shouldertap.dev/api/experts.md): Read the expert registry with live state, or replace it wholesale, over the ShoulderTap HTTP API.
- [Kinds](https://docs.shouldertap.dev/api/kinds.md): A kind determines the schema an expert's answer is captured into. ShoulderTap ships two built-in kinds and lets consumers register their own.
- [Proposals](https://docs.shouldertap.dev/api/proposals.md): List pending proposals and accept or reject them over the ShoulderTap HTTP API — the approval queue.
- [Requests](https://docs.shouldertap.dev/api/requests.md): Submit a context request and read its status and answer over the ShoulderTap HTTP API.
- [Slack events](https://docs.shouldertap.dev/api/slack-events.md): The endpoint that receives Slack Events API callbacks so ShoulderTap can capture expert replies.
- [CLI reference](https://docs.shouldertap.dev/cli.md): Every shtap command and its flags — init, serve, ask, queue, accept, reject, experts, audit, and mcp.
- [Concepts](https://docs.shouldertap.dev/concepts.md): The core building blocks of ShoulderTap — requests, proposals, experts, consumers, kinds, topics, transports, and adapters — and how they fit together.
- [MCP integration](https://docs.shouldertap.dev/consumers/mcp.md): Any MCP-capable agent can tap a human expert with no custom integration, using ShoulderTap's ask_expert and check_answer tools over stdio.
- [Integrating an agent](https://docs.shouldertap.dev/consumers/overview.md): A consumer is anything that asks ShoulderTap for context — an AI agent, a service, or the CLI. Learn how consumers submit requests, declare kinds, and receive answers.
- [Python SDK](https://docs.shouldertap.dev/consumers/sdk.md): Submit requests and receive answers from Python with ShoulderTapClient, or register in-process callbacks for a consumer embedded alongside the engine.
- [Webhook consumers](https://docs.shouldertap.dev/consumers/webhooks.md): Register a webhook consumer to receive ShoulderTap's proposal and request events as HTTP POSTs, and write accepted answers into your system of record.
- [Write-back adapters](https://docs.shouldertap.dev/consumers/write-back.md): Turn an accepted proposal into a write against a system of record using ShoulderTap's OpenMetadata and generic webhook adapters.
- [What is ShoulderTap?](https://docs.shouldertap.dev/introduction.md): ShoulderTap routes an AI agent's unanswerable questions to a human expert on Slack, captures the reply as structured knowledge with provenance, and writes it back after human approval.
- [Quickstart](https://docs.shouldertap.dev/quickstart.md): Run the full ShoulderTap loop in your terminal in about 90 seconds using the console transport — no Slack app or credentials required.
- [Approving answers](https://docs.shouldertap.dev/run/approvals.md): Every expert answer becomes a pending proposal that a human reviews before it reaches a consumer. Approve or reject from the web UI or the CLI.
- [Configuration](https://docs.shouldertap.dev/run/configuration.md): Every shouldertap.yaml setting — organization, LLM, Slack, server, defaults, experts, and topics — plus how secrets map to environment variables.
- [Experts and routing](https://docs.shouldertap.dev/run/experts.md): Define the expert registry, understand how ShoulderTap routes a request to the right person, and control escalation, rate limits, quiet hours, and opt-outs.
- [Install ShoulderTap](https://docs.shouldertap.dev/run/install.md): Install the shouldertap package, generate a starter configuration with shtap init, and understand what gets written to disk.
- [Run the server](https://docs.shouldertap.dev/run/serve.md): Start the ShoulderTap engine with shtap serve, choose a transport, and reach the HTTP API and approval UI.
- [Connect Slack](https://docs.shouldertap.dev/run/slack.md): Set up the Slack transport so ShoulderTap asks experts by direct message and captures their replies through the Slack Events API.