Codabench MCP server
io.github.harshitAgr/codabench-mcp·v0.1.1MCP server for the Codabench REST API — drives a full ML-benchmark participant workflow.
0 stars23 downloads/wk
Reviews
Write oneNobody has reviewed Codabench yet.
If you have run it, two minutes of your experience saves the next person an afternoon.
Codabench tools (16, 1 write)
write = sends, deletes, buys or postsRead from the package source without running it. The installed server may list more.
codabench_requestdownload_datasetget_competitionGet full details of a single Codabench competition.
get_competition_rulesget_leaderboardGet the leaderboard for a phase.
get_my_profileGet the authenticated user's Codabench profile.
get_phaseGet full details of a single phase.
get_submissionGet full details of a submission including scores.
get_submission_logsGet stdout/stderr/ingestion logs for a submission.
get_taskGet full details of a single task, including dataset keys.
list_competition_phasesList all phases of a Codabench competition.
list_competition_tasksList all tasks attached to a Codabench competition.
list_my_submissionsList the authenticated user's submissions, newest first.
poll_submissionsearch_competitionssubmit_to_phasewrite action
Public scan report
scanner v0.1.9 · 2026-09-27 · same rubric, same numbers if you re-run it
- Code scan13 source files scanned25/25
- –Live reliabilityno gateway calls yet and no remote to proben/a
- –Tool poisoningtools not inspected (local package is not executed); not countedn/a
- Auth qualitystatic API keys via environment variables6/15
- Maintenancelast push 118 days ago8/15
- Maintainer identityregistry namespace matches repository owner; GitHub account older than a year8/10
How Codabench compares
Against every graded MCP server, from our own scans and uptime checks. Updated after each scan run.
- Safety score
- 72/100
- higher than 38% of 48,255 servers
- Tools
- 16
- the typical server has 7
- Tools that change things
- 6%
- typical: 0%. Fewer write tools means less an agent can break
- Downloads a week
- 23
- more than 9% of 12,191 servers with a package
What its tools let an agent do: 15 read · 1 change
Install Codabench in Claude Code, Cursor or VS Code
claude mcp add codabench-mcp -- uvx codabench-mcp
What the publisher says
From the Codabench repository's README, as published. We do not edit it. Read it on GitHub
codabench-mcp
MCP server for the Codabench REST API. Lets an AI agent drive a full participant ML-benchmark workflow: discover competitions, read rules, download data, submit, poll, read leaderboards.
Features
Curated tools for the participant path:
getphase, getcompetitionrules, listcompetitiontasks, gettask
- searchcompetitions, getcompetition, listcompetitionphases,
- download_dataset (streaming + SHA-256)
- listmysubmissions, getsubmission, getsubmission_logs
- submittophase (handles Codabench's 3-step upload flow)
- pollsubmission (backoff + timeout + non-error "stillrunning")
- getleaderboard, getmy_profile
- codabench_request — generic REST escape hatch (GET-only by default)
Authentication
You need a Codabench API token. Two ways to get one:
1. Swagger UI (recommended)
- Visit https://www.codabench.org/api/docs/
- Find the api-token-auth endpoint → Try it out
- POST your username/password
- Copy the token value from the response
2. curl
curl -X POST https://www.codabench.org/api/api-token-auth/ \
-H "Content-Type: application/json" \
-d '{"username":"YOUR_USERNAME","password":"YOUR_PASSWORD"}'The response is {"token": "..."}.
Install
uvx codabench-mcpTo run the bleeding-edge main branch instead of the last release:
uvx --from git+https://github.com/harshitAgr/codabench-mcp codabench-mcpOr for development:
git clone https://github.com/harshitAgr/codabench-mcp.git
cd codabench-mcp
uv sync
uv run codabench-mcp # requires CODABENCH_API_TOKENConfigure an MCP client
Claude Desktop (claudedesktopconfig.json)
{
"mcpServers": {
"codabench": {
"command": "uvx",
"args": ["codabench-mcp"],
"env": {
"CODABENCH_API_TOKEN": "paste-your-token-here"
}
}
}
}Claude Code (plugin marketplace)
/plugin marketplace add harshitAgr/codabench-mcp
/plugin install codabench-mcp@codabench-mcpThen export your token in the shell where you launch Claude Code:
export CODABENCH_API_TOKEN=paste-your-token-hereClaude Code (manual)
If you'd rather skip the plugin layer:
claude mcp add codabench \
--env CODABENCH_API_TOKEN=paste-your-token-here \
-- uvx codabench-mcpEnvironment variables
Development
uv sync
uv run pytest # unit tests, no network
uv run ruff check . # lint
uv run ruff format --check . # format checkLicense
MIT — see LICENSE.
mcp-name: io.github.harshitAgr/codabench-mcp
Nothing above is checked by us. What we check is on the safety report.
Codabench: common questions
- Is Codabench MCP server safe?
- Mostly: it is graded B (72/100). Read the Codabench safety report
- How do I install Codabench?
- It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.
- Does Codabench need an API key?
- Yes. The registry entry asks for
CODABENCH_API_TOKEN. - Is Codabench maintained?
- The last commit was 156 days ago (2026-06-01). The latest release is v0.1.1.
- How does Codabench compare with other MCP servers?
- safety score 72/100 (higher than 38% of 48,255 servers); tools 16 (the typical server has 7); tools that change things 6% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.