Turn GitHub into a REST + MCP endpoint

Writ turns the GitHub actions you already do in a browser into a callable REST endpoint your software (and your AI agents) can call.

curl -X POST https://api.usewrit.app/v1/github/run \
  -H "Authorization: Bearer wt_8f2c...a91" \
  -H "Content-Type: application/json" \
  -d '{ "input": { "query": "your search term" } }'

Record the GitHub task once (open a repo, then read issues and PRs) and call it as a REST endpoint and an MCP tool. No backend project, no waiting on GitHub to ship an API.

1. 1
   Record the GitHub task once

   Open the recorder and open a repo on GitHub in a real browser. Writ captures each step, including login and 2FA, into a reusable workflow.

2. 2
   Publish it as an endpoint

   One click turns the workflow into a callable REST endpoint at /v1/github/run and a matching MCP tool. Inputs become a typed schema.

3. 3
   Call it from anything

   Call GitHub from your code, a no-code tool, or an AI agent. Run it on your own machine for free, or on metered cloud.

Read the REST API docs, see what's possible on workflows, or search for another site. Free to start. See pricing.

GitHub has an API, but some org settings, Insights views, and enterprise admin screens are only reachable through the browser UI.

Writ closes that gap. Record the GitHub task once (or describe it to an AI session) and publish it as a REST endpoint at /v1/github/run and a matching MCP tool. Writ runs on your own accounts, with your own credentials and data, on sites you are authorized to use.

Each operation below is a real GitHub action recorded into a workflow and exposed as a typed endpoint.

Returns structured fields such as repo metadata, stars, open issues, PR status: typed JSON, not raw page text.

POST your inputs, get structured output. Authenticate every call with a scoped wt_ key.

curl -X POST https://api.usewrit.app/v1/github/run \
  -H "Authorization: Bearer wt_8f2c...a91" \
  -H "Content-Type: application/json" \
  -d '{ "input": { "query": "your search term" } }'

import requests

r = requests.post(
    "https://api.usewrit.app/v1/github/run",
    headers={"Authorization": "Bearer wt_8f2c...a91"},
    json={"input": {"query": "your search term"}},
)
print(r.json()["output"])

{
  "run_id": "run_gith_7b21",
  "status": "succeeded",
  "output": {
    "repo_metadata": "...",
    "stars": "...",
    "open_issues": "..."
  }
}

The reason GitHub has no usable API is the same reason it's hard to automate. Writ handles it.

Writ signs into your own GitHub account through a Persona with 2FA for UI-only actions, or reads public repos with no login.

Pick whichever fits the task. All three produce the same callable workflow.

Built on the same engine as the recorder and workflows.

Monitor GitHub for new release, new issue matching a label, star threshold and fire this endpoint the instant something changes.

Monitors check as fast as every 10s (10s on Enterprise). A detected change becomes an automation trigger that runs your workflow in seconds. Monitor GitHub ▸

Build a GitHub API

Record it once, then call it from your software and your AI agents. Get started free.