cv-pro

Turn a PDF resume into a living personal site at 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 cv-pro

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 cv.ha7ch.com/{handle}.

Step 2 — Register and log in (one command)

npx cv-pro@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 cv-pro@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 cv-pro@latest whoami

Tell the user their live URL: https://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 cv-pro@latest register <handle>      # claim handle + auto-login (no browser)
npx cv-pro@latest login cv_pat_...       # save an existing token
npx cv-pro@latest whoami                 # show handle + page URL
npx cv-pro@latest get                    # print current resume JSON
npx cv-pro@latest get --variant=openai   # print one stored variant JSON
npx cv-pro@latest update resume.json     # replace full resume
npx cv-pro@latest update-section <section> data.json
npx cv-pro@latest open                   # open live page in browser
npx cv-pro@latest open --json            # open the public JSON view

Public JSON URL

Every resume is also served as raw JSON at https://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://cv.ha7ch.com/api/mcp \
  --header "Authorization: Bearer cv_pat_..."

Cursor — ~/.cursor/mcp.json

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

Codex — ~/.codex/config.json

{
  "mcpServers": {
    "cv": {
      "type": "http",
      "url": "https://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 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: cv.ha7ch.com
• Repo: github.com/LAWTED/cv-pro