# DEPLOY API documentation

Developer reference for DEPLOY's verification framework as a callable surface. The registry publishes its discipline through three complementary channels; consumers compose them per the integration that fits.

## Three surfaces, one corpus

- **[MCP tool catalog](/docs/mcp.md)**: 20 tools across two surfaces. V1 carries five retrieval primitives (entity lookup, deployments by location or model, recent incidents). V2 carries fifteen framework-analysis tools (claim verification, aggregator drift detection, autonomy-boundary and brain-provider-architecture classification, methodology resolution, corrections journal access). The MCP server lives at [/mcp](/mcp); this catalog is the per-tool reference.
- **REST API at /v1/***: HTTP JSON endpoints for every entity type on the registry plus the public taxonomies, sagas, and methodology surfaces. Each entity detail response carries an additive `framework_metadata` block with the same verification posture, cap-flags, and within-cohort pair membership the methodology surface renders. Reference page forthcoming in the next docs batch.
- **JSON-LD with deploy: namespace**: every entity page emits structured data under the schema.org graph plus DEPLOY's custom vocabulary at `https://deploy.report/schema/v1/`. The deploy: properties surface methodology metadata, autonomy-boundary classifications, brain-provider architectures, within-entity sagas, and funding-round graphs. Vocabulary reference page forthcoming.

## Framework reference

The discipline behind every tool, endpoint, and JSON-LD block is documented at four canonical references on the registry:

- [/methodology](/methodology.md): the verified-vs-claimed rubric; what counts as verification.
- [/editorial-process](/editorial-process.md): how the framework operates day-to-day. Lane discipline, audit-first cohort sync, four-tier layered verification.
- [/corrections](/corrections.md): the public log of records changed since publication (append-only; same surface as the v2.list_corrections tool).
- [/funding](/funding.md) and [/conflicts](/conflicts.md): independence and relationship disclosure standing.

## Integration patterns

Three common shapes drive integration choice:

- **Agentic AI assistants**: connect via MCP at /mcp; the tools/list discovery flow returns all 20 tools with input schemas and framework-laden descriptions. Recommended starting point for an LLM that should verify claims before propagating.
- **Application integration and analytics**: query the REST API at /v1/*. Authentication via `Authorization: Bearer <token>` against the public token surface. Every entity detail response carries framework_metadata for downstream verification framing.
- **Structured data ingestion and AEO**: crawl entity pages directly; parse JSON-LD with deploy: namespace vocabulary. The custom vocabulary surfaces verification posture, cap-flags, and architectural classifications that schema.org alone cannot express.

## Machine-readable mirror

Every docs page also ships a Markdown mirror at the same path with a `.md` suffix: [/docs.md](/docs.md), [/docs/mcp.md](/docs/mcp.md). Same content, agent-readable shape. The registry-wide manifest is [/llms.txt](/llms.txt).

_Canonical URL: /docs. HTML version: [/docs](/docs)._