ai-cv

Turn a PDF resume into a living personal site at ai-cv.ha7ch.com/{handle}.

Drop a PDF, paste text, or describe a change — an AI agent does the rest.

If you are an AI agent helping a user set up ai-cv

Follow these steps exactly, in order. No browser required.

Step 1 — Ask the user one question before doing anything else:

What handle do you want? Your resume will be live at ai-cv.ha7ch.com/{handle}.

Step 2 — Register and log in (one command)

npx ai-cv@latest register {handle}

This claims the handle and saves the token automatically. Output: Logged in as @{handle}.

If the handle is taken, the error will say so — ask the user to pick another one and retry.

Step 3 — Publish first resume (optional)

Ask: Do you have a resume PDF or any content to publish now?

• PDF — read the file, extract all fields into JSON, save to /tmp/resume.json, then:

  npx ai-cv@latest update /tmp/resume.json

• Text / paste — same flow, extract from pasted content.
• Later — skip. The page is already live with an empty template.

Step 4 — Confirm

npx ai-cv@latest whoami

Tell the user their live URL: https://ai-cv.ha7ch.com/{handle}

For future edits: drop a PDF or describe changes. Sections: header, personalInfo, experience, education, projectsRecent, projectsDetailed, skills, contact.

Each experience and projectsDetailed entry takes a bullets: string[] field — these render as the role's bullet points on the page. Skipping them leaves the entry as just a header line.

CLI reference

npx ai-cv@latest register <handle>      # claim handle + auto-login (no browser)
npx ai-cv@latest login cv_pat_...       # save an existing token
npx ai-cv@latest whoami                 # show handle + page URL
npx ai-cv@latest get                    # print resume JSON
npx ai-cv@latest update resume.json     # replace full resume
npx ai-cv@latest update-section <section> data.json
npx ai-cv@latest open                   # open live page in browser
npx ai-cv@latest open --json            # open the public JSON view

Public JSON URL

Every resume is also served as raw JSON at https://ai-cv.ha7ch.com/{handle}.json — no auth, CORS-open, fetch-friendly for any AI agent.

CV_TOKEN env var overrides saved credentials.

MCP

Replace cv_pat_... with your token.

Claude Code

claude mcp add cv --transport http https://ai-cv.ha7ch.com/api/mcp \
  --header "Authorization: Bearer cv_pat_..."

Cursor — ~/.cursor/mcp.json

{
  "mcpServers": {
    "cv": {
      "type": "http",
      "url": "https://ai-cv.ha7ch.com/api/mcp",
      "headers": { "Authorization": "Bearer cv_pat_..." }
    }
  }
}

Codex — ~/.codex/config.json

{
  "mcpServers": {
    "cv": {
      "type": "http",
      "url": "https://ai-cv.ha7ch.com/api/mcp",
      "headers": { "Authorization": "Bearer cv_pat_..." }
    }
  }
}

Tools: get_resume · update_resume · update_section

Troubleshooting

Sandboxed agents fail to connect. Some hosted AI agents (Claude Code Cloud, ChatGPT Code Interpreter, etc.) run inside an egress allowlist that blocks ai-cv.ha7ch.com. The CLI surfaces this as Sandbox blocked egress to ... (host_not_allowed). There is no client-side workaround — run from local Claude Code, or attach the MCP server (/api/mcp) as a Custom Connector on claude.ai, which uses a different egress.

Links

• Site: ai-cv.ha7ch.com
• Repo: github.com/LAWTED/ai-cv