PAGA Remote Tool (MCP Server)#
TOU validation and deployment status (2026-08-16)#
Scanpy/PAGA, loopback discovery/call, and invalid-cluster rejection passed on connected deterministic trajectories, including a finite 3 x 3 matrix. A fully disconnected fixture still triggers a sanitized upstream failure. Public publication, cross-user isolation, representative accuracy, broad concurrency, and recovery remain incomplete. Authenticated private Platform import and owner testing passed on 2026-08-16; public publication and independent-caller authorization/isolation remain untested.
Operation:
run_paga_trajectoryStart:
python -m tooluniverse.remote.paga.paga_toolEndpoint:
http://127.0.0.1:8022/mcpProvider configuration: set
TOOLUNIVERSE_REMOTE_DATA_ROOT;cluster_keymust contain bounded valid categories in an approved H5AD input.TOU check:
tu doctor --forward http://127.0.0.1:8022/mcp --jsonPrivate relay:
tu serve --share --forward http://127.0.0.1:8022/mcp --name validation-paga --workers 1
Non-loopback binding requires TOOLUNIVERSE_API_TOKEN; otherwise keep the server on loopback. The relay requires TOOLUNIVERSE_SERVICE_KEY. The disconnected-fixture failure is not counted as a pass.
New-user check: python scripts/remote_validation/setup_skill_preflight.py --implementation paga. Add --check-provider-env before launch, --live after launch, and --check-connect-prereqs before sharing. Live preflight checks exact MCP discovery only; it does not run or validate a model. The pinned relay SDK is not on PyPI and currently requires authorized GitHub repository access plus a configured SSH key, so a working local MCP server does not by itself prove that a new operator can share it.
The authenticated 2026-08-16 Platform matrix found all 30 private owner relays online and all 41 operations discoverable. This implementation was imported as unpublished owner draft(s), configured with a 120-second timeout, and invoked through /expert-sessions/{id}/test. Across the set, 38 unique operations passed return-schema and semantic validation; the three USPTO operations returned exact provider HTTP 403 and remain credential-blocked. Public publication, independent-caller authorization/isolation, broad saturation, and persistent supervision were not tested.
See the complete setup and verification guide.
Serves PAGA
(Partition-based Graph Abstraction; Wolf et al., Genome Biology 2019) as a
ToolUniverse remote tool: run_paga_trajectory — estimates connectivity between
single-cell clusters and reports the strongest cluster pairs (the trajectory
backbone).
Served remotely (not bundled) because PAGA pulls in scanpy + anndata + igraph +
leidenalg. Computation is cheap once a neighbors graph exists; if one is absent
the tool log-normalizes, runs PCA, and builds neighbors before sc.tl.paga.
Deploy#
pip install -r requirements.txt # scanpy + igraph + leidenalg
python paga_tool.py # starts the MCP server on 127.0.0.1:8022
Inputs are referenced by adata_path (a server-accessible .h5ad whose obs
carries the requested cluster_key), since single-cell matrices are large. The
cluster x cluster connectivity matrix returned is small and bounded. Expose
remotely only behind TOOLUNIVERSE_API_TOKEN (SMCP bind guard).
Register in ToolUniverse#
Tool definition: src/tooluniverse/data/remote_tools/paga_tools.json
(type: RemoteTool). Connect via the standard MCPAutoLoaderTool/server_url
mechanism.