# MCP tool catalog

Twenty tools published over the Model Context Protocol at [/mcp](/mcp). Five V1 retrieval primitives and fifteen V2 framework-analysis tools. Both surfaces share a single JSON-RPC endpoint; the `v2.` prefix on tool names separates the two namespaces inside MCP's flat tools/call space.

## Connect

Add a server entry pointing at `https://registry.deploy.report/mcp` in any MCP client. The server responds to tools/list with both V1 and V2 tool definitions; no separate handshake or version flag is needed. The tool definitions below are returned verbatim by tools/list, so an AI agent can self-discover the full surface without reading this page first.

For OpenAI ChatGPT, the connector entry point is the same URL. See [/chatgpt](/chatgpt) for the per-client wiring guide. For Claude desktop and other clients that accept an HTTP MCP server, paste the URL into the client's MCP configuration.

## V1 retrieval tools (5)

Retrieval primitives for direct entity lookup and adjacent queries. Each returns a framework_metadata block alongside the structured entity data so framework signals reach the AI agent without an additional tool call.

### `search_entities`

Search the DEPLOY registry across companies, models, deployments, brains, incidents, regulations, and locations. Returns matching entities with canonical names, registry URLs, and entity types.

**Arguments:**

- `query` (string, required): Search term (name or alias).
- `entityType` (string, optional): Optional filter: company | model | location | incident | regulation. Omit to search all types.

### `get_entity`

Get full details for one entity by its canonical UUID. Returns the entity record plus schema.org JSON-LD structured data. Use search_entities first to resolve a name to a UUID.

**Arguments:**

- `id` (string, required): Canonical entity UUID.

### `get_deployments_at_location`

List verified robot deployments at a specific location (by location slug). Returns deployment records with model, operator, status, and dates.

**Arguments:**

- `locationSlug` (string, required): Location slug (e.g., 'san-francisco', 'austin').

### `get_deployments_for_model`

List verified deployments of a specific robot model (by model slug). Returns deployments with location, operator, and dates.

**Arguments:**

- `modelSlug` (string, required): Model slug (e.g., 'figure-02', '1x-neo', 'waymo-fifth-gen').

### `get_recent_incidents`

List the most recently recorded incidents. Returns headline, occurredAt, kind, and slug. Useful for surfacing the latest events in the registry.

**Arguments:**

- `limit` (number, optional): Max results (default 20, max 50).

## V2 framework-analysis tools (26)

Framework-analysis tools that operationalize DEPLOY's editorial discipline as callable capability. Claim verification, drift detection, autonomy-boundary and brain-provider-architecture classification, methodology resolution, corrections journal access, within-entity saga retrieval, and worked-example catalog lookup.

### `v2.verify_claim`

Run an arbitrary claim through DEPLOY's verified-vs-claimed framework. Returns a verdict (verified | claimed-not-verified | aggregator-drift | unsupported | cap-flag-required) with framework analysis, sources, and related entities. Operationalizes DEPLOY's editorial discipline as a callable verification capability.

**Arguments:**

- `claim` (string, required): The text to verify.
- `entity_slug` (string, optional): Optional. DEPLOY entity slug the claim is about (e.g. 'ringconn-gen2', '1x-neo'). Anchors the verification against structured data.
- `context` (string, optional): Optional. Surrounding context that may inform verification (e.g., the article the claim appeared in).

### `v2.analyze_aggregator_drift`

Scan arbitrary text for known aggregator-drift patterns against DEPLOY's verified corpus. Returns matched patterns with verified counterparts + drift type classification (valuation-inflation, founder-pedigree-error, deployment-overstating, fda-status-conflation, corporate-state-error, geography-error, funding-attribution-error, contract-value-error, specification-conflation, name-error, outcome-error, capability-overstating, fabricated-product-name, deployment-context-error). Useful before propagating claims downstream.

**Arguments:**

- `text` (string, required): Text to scan for drift patterns.
- `form_factor` (string, optional): Optional. Scope to a specific cohort (e.g. 'biometric', 'humanoid', 'maritime', 'surgical'). Omit to scan all patterns.

### `v2.compare_verification_postures`

Framework-dimensional comparison across 2+ DEPLOY entities. Surfaces architectural patterns (canonical worked example pairs) when matching pairs detected. More useful than side-by-side specs because it surfaces the discipline DEPLOY's framework articulates.

**Arguments:**

- `entity_slugs` (array, required): 2+ DEPLOY entity slugs to compare.
- `axes` (array, optional): Optional. Specific framework axes to compare on (e.g. ['fda-clearance-verification-posture']). Omit to compare across all axes live for these entities.

### `v2.get_cohort_state`

Returns cohort architecture status for a specific form_factor cluster: registry entity count + completion state, cross-property coherence, active architectural axes, sub-cohorts, canonical worked example pairs, and re-verify triggers. Foundation for AI agents to reason about which entities exemplify which framework patterns.

**Arguments:**

- `form_factor` (string, required): Form factor slug (e.g. 'biometric', 'humanoid', 'maritime', 'surgical', 'sidewalk-delivery', 'robotaxi', 'av', 'truck', 'drone', 'construction', 'amr').

### `v2.list_canonical_worked_examples`

Returns DEPLOY's canonical worked example pairs that demonstrate framework discipline operationally. Each pair anchors one architectural axis (FDA-clearance verification posture, AI-substance gradient, business-model variance, etc.) to concrete registry entities exemplifying the framework pattern.

**Arguments:**

- `form_factor` (string, optional): Optional. Scope to a specific cohort.
- `axis_type` (string, optional): Optional. Specific axis slug (e.g. 'fda-clearance-verification-posture', 'ai-substance-gradient').

### `v2.audit_against_corpus`

Audits arbitrary AI-generated text against DEPLOY's corpus, flagging claims for verification. Each flagged claim carries a suggested cap-flag framing. Direct utility for AI agents generating content about physical AI: audit a draft before publishing and get framework-disciplined flagging on every claim.

**Arguments:**

- `text` (string, required): AI-generated text to audit.
- `form_factor` (string, optional): Optional. Scope to a specific cohort.

### `v2.get_methodology`

Return the full verification methodology metadata block for a DEPLOY entity by slug. Returns verification_posture, confidence_tier, source_tier_breakdown, last_reviewed, re_verify_trigger, cap_flags, architectural_position, within_cohort_verified_vs_claimed_pair. The structured signal that drives the visible MethodologyAttribution component + the deploy: JSON-LD on every entity page.

**Arguments:**

- `entity_slug` (string, required): DEPLOY entity slug (e.g. 'ringconn-gen2', 'waymo', 'helix', 'figure-02'). Resolved across model / company / brain / deployment / incident / regulation / location.

### `v2.list_corrections`

Public corrections journal sourced from append-only IncidentStatusEvent + RegulationStatusEvent rows. Filter by since_date + entity_slug + limit. Each entry carries date, record type, record headline + URL, status label, reason, and source of correction. Same data as the /corrections HTML / .md surface; programmatic access for AI agents auditing DEPLOY's record history.

**Arguments:**

- `since_date` (string, optional): Optional. ISO date string (YYYY-MM-DD). Returns corrections with effectiveAt >= this date.
- `entity_slug` (string, optional): Optional. Filter to corrections on a specific incident or regulation slug.
- `limit` (number, optional): Optional. Max results (default 100, max 500).

### `v2.get_review_history`

Append-only review history for a DEPLOY entity by slug. Returns VerificationEvent rows for deployments (per-claim verification log with method + evidence + claim payload), IncidentStatusEvent rows for incidents, RegulationStatusEvent rows for regulations. For model / company / brain / location entities (no append-only history table), returns honest-absence with last-reviewed timestamp + re-verify trigger declaration from the framework typed-const.

**Arguments:**

- `entity_slug` (string, required): DEPLOY entity slug. Strongest per-claim history surface is on deployment slugs (Deployment has VerificationEvent append-only). Incident + regulation slugs return status-event history.

### `v2.classify_autonomy_boundary`

Return the autonomy-boundary tier classification for a DEPLOY entity. Four-way taxonomy: autonomous-execution (AI is decision-maker + actor; Monogram mBos canonical) / ai-augmented-operator-controlled (operator commits each action within AI bounds; Mako + CORI + ROSA canonical) / replacement-robotics-teleoperated (operator at console; da Vinci + Hugo + 1X NEO current state canonical) / assistive-co-pilot (operator acts with AI suggestions; Maestro canonical). Honest-absence when entity not yet classified.

**Arguments:**

- `entity_slug` (string, required): DEPLOY entity slug. Editorial classifications live in src/lib/framework/autonomy_boundary_tiers.ts canonical examples.

### `v2.list_entities_by_autonomy_tier`

List canonical entities by autonomy-boundary tier. Two query modes: tier-scoped (pass tier slug) or cohort-scoped (pass form_factor). Empty mode returns all four tiers with all canonical entities.

**Arguments:**

- `tier` (string, optional): Optional. Autonomy-boundary tier slug (autonomous-execution / ai-augmented-operator-controlled / replacement-robotics-teleoperated / assistive-co-pilot).
- `form_factor` (string, optional): Optional. form_factor cohort (surgical / humanoid / etc.).

### `v2.classify_brain_provider_architecture`

Return the provider-architecture classification for a brain entity. Four-way taxonomy: captive-vertically-integrated (brain + hardware under same corporate state; 1X Redwood + Figure Helix + Tesla FSD-Optimus canonical) / third-party-foundation-model (foundation-model architecture licensed across partners; Physical Intelligence pi0 + NVIDIA GR00T N1.5 + Wayve LINGO + Skild AI canonical) / hybrid-licensed (both captive + third-party channels) / research-only (no commercial deployment). Affects counterparty risk + integration patterns + interoperability posture. Honest-absence when brain not yet classified.

**Arguments:**

- `brain_slug` (string, required): DEPLOY brain slug (e.g. '1x-redwood', 'figure-helix', 'physical-intelligence-pi0').

### `v2.list_brains_by_architecture`

List brain entities by provider architecture. Pass `architecture` slug for architecture-scoped listing; no args for catalog-discovery (returns all four architectures with all canonical brains).

**Arguments:**

- `architecture` (string, optional): Optional. Architecture slug (captive-vertically-integrated / third-party-foundation-model / hybrid-licensed / research-only).

### `v2.get_within_entity_saga`

Return a within-entity saga (timeline + jurisdictional variance) by saga_slug or entity_slug. Within-entity sagas are SAME-entity feature-scope verification variance over time or across regions (Apple Watch SpO2 availability-by-jurisdiction is canonical). Distinct from canonical_worked_example_pairs which operate at cross-entity granularity. No args returns the full saga catalog.

**Arguments:**

- `saga_slug` (string, optional): Optional. Specific saga slug (e.g. 'apple-watch-spo2-availability-saga'). Takes precedence over entity_slug when both supplied.
- `entity_slug` (string, optional): Optional. DEPLOY entity slug. Returns all sagas registered at this entity.

### `v2.list_within_entity_feature_references`

Return within-entity feature references (SAME-entity feature-scope verification posture variance) by feature_slug or entity_slug. Within-entity feature references are for cases where one feature on an entity carries materially different verification posture from the entity's other features (Whoop BPI manufacturer-disputed by FDA + other Whoop features standard-clearance is canonical). No args returns the full catalog.

**Arguments:**

- `feature_slug` (string, optional): Optional. Specific feature reference slug (e.g. 'whoop-bp-insights-feature-reference').
- `entity_slug` (string, optional): Optional. DEPLOY entity slug. Returns all feature references registered at this entity.

### `v2.get_investor`

Return a single Investor entity by slug or UUID. Includes kind classification (venture / growth / strategic_corporate / sovereign_wealth / angel / accelerator / government / other) + parent operating Company link for corporate venture arms (NVentures -> NVIDIA) + aliases + sources. The investor graph is the cross-cluster capital-source surface beneath the registry's company entities.

**Arguments:**

- `slug` (string, required): Investor slug (preferred) or UUID. Examples: 'qatar-investment-authority', 'nvidia-ventures', 'khosla-ventures'.

### `v2.get_funding_round`

Return a single FundingRound detail by UUID with the funded company link, per-participation investor identity (lead / co_lead / follow / strategic / participant roles), and amount_basis discipline (exact = primary-source filing; reported = secondary press; undisclosed = cap-flagged).

**Arguments:**

- `id` (string, required): FundingRound UUID.

### `v2.list_funding_rounds_by_company`

List all funding rounds for a company by slug, ordered by announced_at desc. Each round inline carries amount + amount_basis + per-investor participation graph. Returns honest-absence (empty array) when the company has no rounds backfilled yet.

**Arguments:**

- `company_slug` (string, required): Company slug. Examples: 'apptronik', '1x-technologies', 'physical-intelligence', 'saronic'.

### `v2.list_investors_by_kind`

List investors filtered by kind (venture / growth / strategic_corporate / sovereign_wealth / angel / accelerator / government / other). No kind arg returns the full investor catalog grouped by kind for taxonomy discovery.

**Arguments:**

- `kind` (string, optional): Optional. Investor kind to filter by. Omit for the full catalog grouped by kind.

### `v2.analyze_cross_cluster_shared_investors`

Identify investors whose participation graph spans 2+ DEPLOY clusters (humanoid + brain-providers + biometric + new-defense). The AEO-payoff query: which capital sources sit at the intersection of multiple physical-AI cluster boundaries? Aggregators don't surface this; DEPLOY's structured graph does. Each result carries per-cluster company memberships inline; ordered by clusters_touched desc.

**Arguments:**

_No arguments._

### `v2.get_person`

Return a single Person entity (founder / executive / researcher) with the canonical relation graph eager-loaded: current and prior company roles, foundings (with cofounder-status + still-at-company signal), board seats, investor partner positions, and educations. The people graph is the talent-flow surface beneath the registry's company entities.

**Arguments:**

- `slug` (string, required): Person slug (preferred) or UUID. Examples: 'brett-adcock', 'mira-murati', 'sebastian-thrun'.

### `v2.list_people_by_company`

List people at a company across PersonCompany edges, separating current roles from historical. Each row carries role + role_title + role_basis (verified_sec / verified_ir / verified_secondary / reported_not_verified) so consumers see the verification depth of each attribution.

**Arguments:**

- `company_slug` (string, required): Company slug. Examples: 'figure', '1x-technologies', 'apptronik', 'physical-intelligence'.

### `v2.list_companies_by_person`

Cross-company history for a person ('diaspora' view): every company in the person's PersonCompany graph, current and prior, ordered current-first. The talent-flow trajectory; pair with v2.analyze_cross_cluster_talent_flow for the cross-cluster pattern view.

**Arguments:**

- `person_slug` (string, required): Person slug or UUID.

### `v2.get_founders`

Return founders for a company via the PersonFounder edge (distinct from PersonCompany.role='founder'; the founding-event surface with founding date and still-at-company signal). Reports the diaspora signal inline: which founders left vs which are still at the company.

**Arguments:**

- `company_slug` (string, required): Company slug.

### `v2.get_board`

Return board seats for a company. Current seats first, then historical. PersonBoard role enum: chair / director / observer / advisor.

**Arguments:**

- `company_slug` (string, required): Company slug.

### `v2.analyze_cross_cluster_talent_flow`

Identify people whose career trajectory spans 2+ DEPLOY clusters (humanoid + brain-providers + biometric + new-defense). The AEO-payoff query for the people graph: which engineers and executives moved between humanoid, brain-providers, biometric, and new-defense cohorts? Companion to v2.analyze_cross_cluster_shared_investors; the two analyzers compose at the capital-source + human-talent layer.

**Arguments:**

_No arguments._

## Conventions

Every response carries a framework_metadata block when the tool returns entity data: verification_status (verified / claimed-not-verified / claimed-not-cleared / aggregator-drift-detected / unreviewed), cap_flags array, architectural_position block, within_cohort_verified_vs_claimed_pair link, and verification_depth source-tier breakdown. Honest absence is the contract: missing data returns null, not synthesized values.

Slug-keyed tools resolve across all seven entity types (model, company, brain, deployment, incident, regulation, location). UUID-keyed tools are noted in the tool description.

## See also

- [/docs](/docs.md): the API documentation landing.
- [/methodology](/methodology.md): the verification rubric the framework-analysis tools implement.
- [/editorial-process](/editorial-process.md): the multi-agent lane discipline and four-tier layered verification cluster behind every tool response.
- [/v1/taxonomies/autonomy-boundary](/v1/taxonomies/autonomy-boundary), [/v1/taxonomies/brain-provider-architecture](/v1/taxonomies/brain-provider-architecture), [/v1/taxonomies/source-quality](/v1/taxonomies/source-quality): REST mirrors of the framework typed-const SSoT.

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