autoapply-mcp

An MCP server that turns a verified candidate profile into a governed job-application pipeline: it discovers postings from public ATS boards, applies hard eligibility gates, scores what survives with quotable evidence, drafts only the answers your profile actually supports, and refuses to submit anything without a recorded human approval.

It is built for high-volume search without becoming a spam bot. Discovery, ranking and form-filling are automated. Judgement, truthfulness and consent are not.

What it does

Related MCP server: job-search

Design rules

1. Nothing is invented. Every drafted answer traces to a profile field or a pre-approved answer. Anything else is handed back to you.

2. Sensitive questions always stop. Work authorization, citizenship, compensation, criminal history, demographics and legal attestations require a human decision by default.

3. Approval binds to content. approve_application records a packet hash. Editing anything invalidates it and submission is refused.

4. Job descriptions are untrusted data. Text from employers is scanned for prompt-injection patterns, wrapped in an explicit data boundary, and never treated as instructions.

5. Anti-bot controls are respected. A detected CAPTCHA aborts the run and hands the application back to you. There is no solving, evading or fingerprint spoofing.

6. Destinations are allowlisted. Submissions only go to hosts the campaign explicitly trusts, over HTTPS.

Requirements

• Node.js 22.5 or newer (uses the built-in node:sqlite, so there is no native build step)

• Playwright, only if you want assisted or automatic form filling: npm install playwright && npx playwright install chromium

Install

git clone <your-fork> autoapply-mcp
cd autoapply-mcp
npm install
npm run build
npm test

Configure

Personal data lives outside the repository, so a checkout can be published as-is. The default home is ~/.autoapply:

mkdir -p ~/.autoapply
cp examples/profile.example.json   ~/.autoapply/profile.json
cp examples/campaign.example.json  ~/.autoapply/campaign.json
cp presets/ai-security-us-canada.json ~/.autoapply/companies.json

Then edit them and verify:

node dist/cli/doctor.js --probe

doctor validates all three files, checks that your resume files exist and are real PDFs, confirms every track points at a resume variant, and probes each configured board.

presets/ai-security-us-canada.json ships 80 company boards that were verified live against the ATS APIs, weighted toward AI, security and infrastructure companies in the US and Canada.

Nothing in this repository is specific to one candidate. Keep profile.json, resumes and the database in ~/.autoapply and they can never be committed by accident.

Environment variables

Register with an MCP client

{
  "mcpServers": {
    "autoapply": {
      "command": "node",
      "args": ["/absolute/path/to/autoapply-mcp/dist/index.js"],
      "env": { "AUTOAPPLY_HOME": "/absolute/path/to/your/.autoapply" }
    }
  }
}

The server speaks stdio. All logs go to stderr, so stdout stays clean for the protocol. AUTOAPPLY_HOME can be omitted if you use the default ~/.autoapply.

Typical session

Single application:

discover_jobs                  -> fetch every board, gate, score, store
list_queue                     -> the ranked, de-duplicated shortlist
explain_job        jobId       -> gate result and scoring evidence
prepare_application jobId      -> match report, employer questions, drafted answers
set_application_content        -> your cover letter and any answers only you can give
preview_application            -> the exact packet plus its hash
approve_application  + hash    -> your explicit authorization
submit_application   mode      -> manual, assisted or auto
record_outcome                 -> track what happened

High volume:

prepare_batch  {tiers:["A","B"], locationClasses:["bay-area"], minCompensation:250000}
preview_batch                  -> the set, plus grouped blocking questions
approve_batch  + manifestHash + expectedCount
submit_batch   mode            -> honours daily limit and pacing
list_batches                   -> progress

See docs/TOOLS.md for every tool, docs/BATCH.md for high-volume campaigns and cached personal data, docs/CONFIGURATION.md for the schemas, docs/RESUMES.md for resume variants and LaTeX builds, and docs/SAFETY.md for the guarantees and their limits.

Submission modes

auto additionally requires the company to appear in submission.allowedCompanies, every required field to be fillable, and no blocked questions to remain unanswered.

What it deliberately does not do

• No LinkedIn, Indeed or Wellfound automation. Their terms prohibit it and their anti-bot systems are built to stop it. Use their alerts as leads, then apply through the employer's own board.

• No Workday support. Each tenant is bespoke, session-bound and protected; a candidate-side integration cannot be done reliably or respectfully.

• No CAPTCHA solving, proxy rotation or fingerprint evasion.

• No writing of your resume prose or cover letters. The server supplies structured evidence; your agent writes the words and you approve them.

Known limitations

• Compensation parsed from description text is a heuristic. Postings that list several geographic pay zones can yield the wrong figure, so those results carry a compensation-parsed-from-text flag. Structured ATS pay data is preferred whenever a board publishes it.

• FX rates in campaign.json are static. Update them before relying on a cross-currency floor.

• Only Greenhouse publishes an application question schema. For Lever and Ashby the server drafts against a baseline field set and reads the real form during an assisted run.

• @modelcontextprotocol/sdk currently pulls a transitive advisory in @hono/node-server affecting its static-file server. This server uses the stdio transport only and never serves static files, so the affected code path is not reachable.

Development

npm run typecheck
npm test
npm run coverage

174 tests cover location classification, compensation parsing, every gate, scoring, the answer policy, resume validation, submission guards, packet hashing, persistence, the ATS adapters and an end-to-end run through a real in-memory MCP client.

License

MIT