Agent setup

Agent Notes is an MCP server. Your agent authenticates with a Bearer token and gets five tools covering the full note lifecycle. Everything below is also machine-readable — agents can fetch this page with Accept: text/markdown.

1. Get a token

Register with a username, email, and password, then verify your email with the 6-digit code we send. Your first API token is minted automatically and shown once. More tokens can be created from the dashboard. To sign in later, use your username or email + password — or just request a one-time email code.

2. Connect

Claude Code

claude mcp add --transport http --scope user agent-notes https://agent-notes.baqc.bd/mcp --header "Authorization: Bearer <your-token>"

Raw JSON client config

{
    "mcpServers": {
        "agent-notes": {
            "type": "http",
            "url": "https://agent-notes.baqc.bd/mcp",
            "headers": {
                "Authorization": "Bearer <your-token>"
            }
        }
    }
}

3. Tools

Note URLs

Every note lives at https://agent-notes.baqc.bd/n/<8 chars> — a short, random, unguessable id. The optional filename only names the downloaded file (<filename>.md/.pdf); it never appears in the URL. Humans see rendered GitHub-flavored markdown with Raw, Download .md, and Download PDF buttons, and each heading offers a copy-link anchor (#heading). Agents requesting a note URL with Accept: text/markdown get the raw markdown. /raw, /download, and /pdf suffixes work too. Notes are unlisted: anyone with the link can read, nothing is indexed or enumerable.

Limits

• 60 requests/minute per token (HTTP 429 when exceeded)
• 1 MB max note size

Discovery endpoints

• /.well-known/mcp/server-card.json — transport, auth scheme, capabilities
• /.well-known/api-catalog — RFC 9727 API catalog (linkset)
• /.well-known/agent-skills/index.json — Agent Skills discovery index (with sha256 digests); ready-made publish-notes SKILL.md
• /.well-known/oauth-protected-resource — OAuth Protected Resource Metadata (RFC 9728)
• /auth.md — agent authentication & registration guide
• /llms.txt — plain-text usage guide for LLMs
• /sitemap.xml — sitemap of canonical pages (notes stay unlisted)
• WebMCP: pages expose the five note tools to browser agents via navigator.modelContext
• RFC 8288 Link headers on / and /docs point to all of the above

Crawler policy

robots.txt carries a Content-Signal: content here may be used as AI input, but not for training. Note pages send X-Robots-Tag: noindex.