Anchord MCP
Type: full-code · Vendor: Anchord · Language: TypeScript · License: proprietary · Status: active · Status in practice: experimental · First released: 2025-01-01
Anchord MCP is an MCP server that gives agents access to the Anchord identity-resolution API so they resolve companies and people to canonical AnchorIDs and run a pre-write safety check before any record write.
Description. Anchord MCP is an MCP server that gives AI agents access to the Anchord identity-resolution API as a set of tools. The resolve tools map companies and people to canonical AnchorIDs against Anchord's identity graph and return needs_review when multiple plausible matches exist. The guard_write tools provide an evaluation-only pre-write safety check that returns allowed true or false with block codes before any record is written. The server is a thin proxy to the hosted Anchord SaaS platform, carrying no business logic of its own.
Agent loop shape. An agent calls resolve_company or resolve_person to map an entity to a canonical AnchorID; if the server returns needs_review the agent stops and surfaces the candidates rather than writing. Before a mutating write the agent calls guard_write, which evaluates the proposed write and returns allowed true or false with block codes. The agent can fetch the merged golden record for an AnchorID at any point. All processing happens server-side; the local MCP server is a thin proxy.
Primary use cases
- resolving companies and people to canonical AnchorIDs
- pre-write safety checks before mutating records
- exporting merged golden records for an entity
- grounding agent-emitted identifiers against a system of record
Key concepts
- AnchorID → canonical-entity-grounding (docs) — The canonical identifier Anchord assigns to a resolved company or person in its identity graph, used as the system-of-record key an agent grounds against instead of inventing its own identifiers.
- needs_review → canonical-entity-grounding (docs) — The status the resolve tools return when multiple plausible matches exist and Anchord cannot auto-resolve with confidence; the agent must stop and surface candidates rather than write.
- guard_write → dry-run-harness (docs) — An evaluation-only pre-write tool that judges a proposed record write and returns allowed true or false plus machine-readable block codes before anything is committed.
- Golden record (docs) — The merged, canonical view of an entity that Anchord exports for a given AnchorID, consolidating attributes from across sources into one authoritative record.
Patterns this full-code implements —
- ★Canonical-Entity Grounding
Anchord's MCP server makes agents resolve companies and people to a canonical AnchorID against its identity graph (system of record) instead of emitting identifiers themselves, returning needs_review…
- ★★Model Context Protocol
Anchord exposes its identity-resolution API to agents through the standard Model Context Protocol tool interface, so any MCP-capable client can call resolve, guard_write, and golden-record tools with…
- ★Dry-Run Harness
The guard_write tools run an evaluation-only pre-write safety check that returns allowed true or false with block codes, so the agent learns whether a proposed write would be accepted before committi…
Neighbourhood
Click any neighbour to follow the lineage. Scroll to zoom, drag to pan.
Alternatives & relatives
- app · closed product
K2view GenAI Data Fusion
K2view assembles a unified per-entity data view for agents; Anchord focuses on resolving entities to canonical IDs and guarding writes, a narrower slice of the same master-data concern.
- full-code · framework
Stainless (API MCP server architecture)
Both wrap a hosted API behind an MCP server as a thin proxy so agents call it as standard tools; Stainless generates MCP servers from API specs while Anchord exposes one specific identity-resolution…
References
Provenance
- Last analyzed:
- Last updated:
- Verification status: partial