interloper-toolkit

The read-only tool functions shared by interloper's AI surfaces — the ADK chat agent (interloper-agent) and the MCP server (interloper-mcp).

Every function takes a frozen ToolkitContext(store, catalog, org_id) as its first argument and returns <SuccessModel> | ToolError — typed pydantic results (see models.py) discriminated by the literal status field, never raising. The models are the tool contract: MCP derives per-tool output schemas from them, row-projecting models act as an allowlist of what leaves the platform, and full-row payloads embed the interloper-db models to keep that coupling visible rather than duplicated. The docstrings are LLM-facing: both consumers surface them verbatim as tool descriptions.

Modules:

• catalog — the component definitions the platform ships (list, detail, asset schemas, field search, schema comparison)
• collection — the org's component instances (read-only listing; sensitive kinds project identity/metadata only)
• lineage — dependency analysis, impact assessment, DAG traversal
• scheduling — jobs, runs, and backfills: read-only monitoring
• analytics — run statistics, partition coverage, data freshness

Mutating operations (trigger/toggle/create) deliberately live with the agent, not here — this package is safe to expose on read-only surfaces.

Depends only on interloper-core and interloper-db; no LLM-framework dependencies (no google-adk, no mcp).