PlanetGraph LLM Discovery Index
Version: 2026-07-31

PlanetGraph provides knowledge graphs for AI agents.

Agents can use PlanetGraph to embed, search, index, discover, and share structured and unstructured data through REST and MCP.

Example use cases include:
- people/company relationship intelligence
- warm introduction discovery
- PlanetGraph Login+ for app sign-up/login and graph-backed user context
- knowledge graph exploration
- graph-backed DataSet catalog discovery
- proprietary graph-backed agent workflows

The people/company relationship dataset is one example dataset and use case, not the definition of the platform.

Primary human docs:
- https://planetgraph.ai/docs
- https://api.planetgraph.ai/docs

Machine-oriented docs:
- https://planetgraph.ai/llms-full.txt
- https://planetgraph.ai/llms/quickstart.txt
- https://planetgraph.ai/llms/recipes-warm-intros.txt
- https://planetgraph.ai/llms/login-plus.txt
- https://planetgraph.ai/llms/api-auth-and-errors.txt
- https://planetgraph.ai/llms/schema-and-query-shapes.txt
- https://planetgraph.ai/llms/bulk-import.txt

Start here for agents:
1) Read quickstart and auth docs.
2) Use OpenAPI at https://api.planetgraph.ai/openapi.json.
3) For a concrete example workflow, use recipes-warm-intros.
4) For app authentication and graph-backed user context, use login-plus.
5) For asynchronous graph ingestion, use bulk-import.

MCP search workflow:
- Call planetgraph_datasets_list to discover curated DataSets, then planetgraph_datasets_describe with a DataSet ID for its canonical Node/Edge types and policy declarations.
- Call planetgraph_schema_describe with optional node_type, edge_type, property_name, or owned_only filters to discover visible types and declared Property capabilities.
- properties entries describe returnability declared by PropertyAccess; searchable_properties entries independently describe Search declarations. Neither implies the other.
- Schema discovery does not inspect Property rows. A declared Property may be absent from an entity, and undeclared Properties may exist.
- Check the truncated object. Narrow by entity type or Property name when a category is truncated.
- Call planetgraph_properties_search for direct lookups. Every searchable property supports exact, prefix, and contains matching, case-insensitively.
- Example: property_name=name, node_type=Company, query=Apple, match_mode=exact.
- Example: property_name=name, node_type=Institution, query=stanford, match_mode=contains finds Stanford University.
- Use include_properties=true with planetgraph_nodes_list or planetgraph_edges_list to inspect a bounded sample of policy-authorized values; property_limit is capped at 25 per entity.
- Prefer property search over Cypher for direct lookups. Use query tools when relationship traversal or broader graph context is required.

Auth summary:
- Interactive docs and API support X-API-Key auth.
- Some client flows also use Firebase Bearer tokens.

Bulk import summary:
- POST /api/v1/imports submits a small inline JSON manifest.
- Large imports use POST /api/v1/imports/uploads, a signed JSONL PUT, then finalize.
- Poll job status and page through results or errors; imports preserve normal ownership, billing, encryption, access, search, projection, and embedding behavior.

Notes:
- /api/v1/edges?type filters edge labels (for example worked_for), not node types.
- Node-type edge filtering is available via source_type and target_type.