harshitagr avatar

Codabench Mcp

harshitagr/codabench-mcp

Wraps the Codabench REST API so Claude can participate in ML competitions end to end. You get tools to search competitions, read rules and task specs, download datasets with streaming and SHA-256 verification, submit solutions through Codabench's three-step upload flow, poll submission status with backoff, and check leaderboards. Authentication uses a Codabench API token you grab from their Swagger UI or via curl. The generic codabench_request tool is locked to GET by default but you can flip CODABENCH_ALLOW_WRITE_RAW if you need it. Useful when you want Claude to explore benchmarks, prototype submissions, or iterate on a competition without leaving the chat.

tools
16
auth
Secret needed
transport
STDIO
env var
1
last updated
Apr 23, 2026

Tools

16 tools

Public tool metadata for what this MCP can expose to an agent.

search_competitions

No parameter schema in public metadata yet.

get_competition

No parameter schema in public metadata yet.

list_competition_phases

No parameter schema in public metadata yet.

get_phase

No parameter schema in public metadata yet.

get_competition_rules

No parameter schema in public metadata yet.

list_competition_tasks

No parameter schema in public metadata yet.

get_task

No parameter schema in public metadata yet.

download_datasetDownloads a dataset with streaming and SHA-256 verification.

Downloads a dataset with streaming and SHA-256 verification.

No parameter schema in public metadata yet.

list_my_submissions

No parameter schema in public metadata yet.

get_submission

No parameter schema in public metadata yet.

get_submission_logs

No parameter schema in public metadata yet.

submit_to_phaseSubmits to a phase, handling Codabench's 3-step upload flow.

Submits to a phase, handling Codabench's 3-step upload flow.

No parameter schema in public metadata yet.

poll_submissionPolls a submission with backoff and timeout, returning a non-error still_running status.

Polls a submission with backoff and timeout, returning a non-error still_running status.

No parameter schema in public metadata yet.

get_leaderboard

No parameter schema in public metadata yet.

get_my_profile

No parameter schema in public metadata yet.

codabench_requestGeneric REST escape hatch, GET-only by default.

Generic REST escape hatch, GET-only by default.

No parameter schema in public metadata yet.

Sponsors

Configuration

CODABENCH_API_TOKEN*secret

DRF token for Codabench. Obtain via https://www.codabench.org/api/docs/ > api-token-auth > Try it out.

Readme

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:

  • search_competitions, get_competition, list_competition_phases, get_phase, get_competition_rules, list_competition_tasks, get_task
  • download_dataset (streaming + SHA-256)
  • list_my_submissions, get_submission, get_submission_logs
  • submit_to_phase (handles Codabench's 3-step upload flow)
  • poll_submission (backoff + timeout + non-error "still_running")
  • get_leaderboard, get_my_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)

  1. Visit https://www.codabench.org/api/docs/
  2. Find the api-token-auth endpoint → Try it out
  3. POST your username/password
  4. Copy the token value from the response

2. curl


View the full README on GitHub

Details

Repository
github.com/harshitagr/codabench-mcp
Category
Automation & Workflows
Package
codabench-mcp
Transport
STDIO
Auth
Required
Registry
MCP registry, active

Related Automation & Workflows MCP servers

View all
1aguantar avatarClickhouse Dataops Mcpaguantar/clickhouse-mcp-serverDataOps ClickHouse MCP server with query optimization and pipeline monitoring0stars
2alexboissav avatarArtefact Revenue Intelligencealexboissav/artefact-mcp-serverRevenue intelligence MCP: RFM analysis, 14.5-point ICP scoring, pipeline health. HubSpot.0stars
3eranimore avatarSurf Session Finderio.github.eranimore/surf-session-finderSurf park sessions schedule availability and notifications for wave park surfers worldwide0stars
4ethanackerman-git avatarPreflight Ios Mcpethanackerman-git/preflight57 tools for AI-powered iOS Simulator automation — tap, swipe, type, screenshots, and more.0stars
5hycore220 avatarK-Work Trusthycore220/k-work-trust-public-demoKorean business record validation and workflow safety gates for AI agents.0stars
6jamesrez avatarValoriajamesrez/valoria-intelligenceMake money with x402. See what's earning, find profitable gaps, and price your service right.0stars