togomcpCAUTION
MCP server for life-science knowledge graphs
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
An MCP (Model Context Protocol) server that gives AI assistants (Claude, etc.) access to biological and biomedical RDF databases via SPARQL at the RDF Portal, as well as selected REST APIs (NCBI E-utilities, UniProt, ChEMBL, PDB, Reactome, Rhea, MeSH, and more).
Quick Start: Remote Server (No Installation)
You can use the hosted TogoMCP server directly — no local setup needed. See https://togomcp.rdfportal.org/ for connection instructions.
Local Installation
Prerequisites
- Python >= 3.11
- uv package manager
1. Install uv
# macOS/Linux curl -LsSf https://astral.sh/uv/install.sh | sh # Windows (PowerShell) powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
2. Clone and install
git clone https://github.com/dbcls/togomcp.git cd togomcp uv sync
3. Set NCBI API Key (required for NCBI tools)
Obtain your NCBI API key and export it:
export NCBI_API_KEY="your-key-here"
Configuration
Claude Desktop
Edit your Claude Desktop config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
~\AppData\Roaming\Claude\claude_desktop_config.json
{
"mcpServers": {
"togomcp": {
"command": "/path/to/uv",
"args": [
"--directory",
"/path/to/togomcp",
"run",
"togo-mcp-local"
],
"env": {
"NCBI_API_KEY": "your-key-here"
}
}
}
}Tip: Runwhich uv(macOS/Linux) orwhere uv(Windows) to find the full path
6933741f07e7OBSERVED · 2026-10-10Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add togo-mcp --env ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} --env ANTHROPIC_AUTH_TOKEN=${ANTHROPIC_AUTH_TOKEN} --env NCBI_API_KEY=${NCBI_API_KEY} --env TOGOMCP_STATS_PASSWORD=${TOGOMCP_STATS_PASSWORD} -- uvx togo-mcp{
"mcpServers": {
"togo-mcp": {
"command": "uvx",
"args": [
"togo-mcp"
],
"env": {
"ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
"ANTHROPIC_AUTH_TOKEN": "${ANTHROPIC_AUTH_TOKEN}",
"NCBI_API_KEY": "${NCBI_API_KEY}",
"TOGOMCP_STATS_PASSWORD": "${TOGOMCP_STATS_PASSWORD}"
}
}
}
}Exposed tools (5)
5 read · 0 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
TogoMCP_Usage_Guide | read | |
get_compound_attributes_from_pubchem | read | |
get_pubchem_compound_id | read | |
get_sparql_endpoints | read | Get the available SPARQL endpoints for RDF Portal. |
mie-refresh | read | Batch-build/revise MIE files (one mie-builder agent per database), then independently re-validate each against the live endpoint |
Trust audit
CAUTIONgrade B · trust 80/100 Install with care. The audit found things worth knowing before you trust its output.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | WARN |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (2 observation(s))
- Network
- declared (9 observation(s))
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (25)
.claude/skills/disease-analysis
.claude/skills/prism
.claude/skills/research-article-analysis
print(f" ! STALE waiver {w['question']} / {w['database']} / {w['token']!r} "f"http://127.0.0.1:{port}/mcp", headers={"Accept": "text/event-stream"}), timeout=5)url = PRODUCTION_URL if args.target == "production" else f"http://127.0.0.1:{args.port}/mcp"code=$(curl -s -o /dev/null -w '%{http_code}' -H "Host: $host" "http://127.0.0.1:${port}/" 2>/dev/null || true)help="question YAML paths (default: all ../../questions/question_*.yaml)")
await pathway_neighborhood(pathway=MAP, seeds="../../get/hsa:207")
"bad", ["../SKILL.md", "/etc/passwd", "references/../../x.md", "references/x.py", "C:\\x.md"]
url = f"http://127.0.0.1:{port}/mcp"servers["togomcp"] = {"type": "http", "url": f"http://127.0.0.1:{port}/mcp"}decoded = _base64.b64decode(header[6:]).decode("utf-8")hdr += f" {'Δsparql':>8s} {'±95%CI':>7s} {'Δcost$':>7s}"hdr += f" {'Δcorr':>7s}"f"Δvs {REFERENCE}: {_fmt(delta, '+.2f')}")<llm-meta-widget llm-url="https://hub.aibranch.org" tool-hub-url="https://hub.aibranch.org" model="qwen3-8-27b-fast" hub-tools="TogoMCP" default-hub-tools="TogoMCP" auth-required="false" enable-model-
**Not an RDF Portal database — KEGG.** `database="kegg"` is invalid on `run_sparql` and `get_MIE_file`: KEGG has no SPARQL endpoint and no MIE. The `kegg_*` tools exist only on a local stdio server wh
**Check your tool list before offering KEGG to the user.** KEGG is OFF BY DEFAULT. The `kegg_*` tools appear only on the local stdio server (`togo-mcp-local`) AND only when the operator sets `TOGOMCP_
- **Antigravity CLI** ([config docs](https://antigravity.google/docs/mcp)) — the tab formerly labelled "Gemini CLI". **Gemini CLI was retired 2026-06-18** for the free tier, Google AI Pro, Ultra and G
Google Cloud access, and paid Gemini API keys. The page had carried the old instructions for six
`add_llm_evaluation.py` (the Opus judge) uses the plain `anthropic` SDK instead, which **requires** `ANTHROPIC_API_KEY` (or `ANTHROPIC_AUTH_TOKEN` / an `ant auth login` profile) — it does not read the
<p><strong>Gemini CLI was retired on 18 June 2026</strong> for the free tier, Google AI Pro, Ultra and Google One, and replaced by <strong>Antigravity CLI</strong>. If that is you, use the Antigravity
<p><strong>Gemini CLI was retired on 18 June 2026</strong> for the free tier, Google AI Pro, Ultra and Google One, and replaced by <strong>Antigravity CLI</strong>. If that is you, use the Antigravity
curl -LsSf https://astral.sh/uv/install.sh | sh
Gates applied: no_behavioural_pass.
6933741f07e7full audit observations/trust-audit/mcp-server/dbcls__togomcp.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-10 | 6933741f07e7 | CAUTION | B | 80 | first audit |
Questions
What is the togomcp MCP server?
MCP server for life-science knowledge graphs
What tools does togomcp expose?
5 in total: 5 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is togomcp safe to connect to an agent?
With care. The audit graded it B (80/100) and found 25 things worth knowing before you trust this server, listed below with the exact line each was found on.
What credentials does togomcp need?
It reads ANTHROPIC_API_KEY, ANTHROPIC_AUTH_TOKEN, NCBI_API_KEY and TOGOMCP_STATS_PASSWORD from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does togomcp run?
It speaks stdio and streamable-http, so it runs as a local process your client starts. It is published on PyPI as togo-mcp.
How current is this page?
The grade is for one exact copy of the source (6933741f07e7), read on 2026-10-10. The repository is watched and re-audited when it changes.