Corpus

Access a user's Corpus library from OpenClaw. Use when the user asks to search saved content, fetch item details, save links into Corpus, or create reminders from Corpus content.

Corpus Skill

Use this skill to read and write a user's Corpus data through the Corpus API.

Required environment variables

• CORPUS_API_TOKEN: user token for Corpus API access.

Generate CORPUS_API_TOKEN

1. Install Corpus AI for iPhone: https://apps.apple.com/us/app/corpus-ai/id6748364607
2. Open the app and sign in.
3. Go to Integrations -> OpenClaw.
4. Create an API token and copy it (the full value is shown once).
5. Set that value as CORPUS_API_TOKEN in your OpenClaw skill env.

Optional environment variables

• CORPUS_API_BASE_URL: API base URL (default: https://corpusai.app).
• CORPUS_TIMEOUT_SECONDS: HTTP timeout in seconds (default: 30).

OpenClaw config example

json

{
  "skills": {
    "entries": {
      "corpus": {
        "path": "/absolute/path/to/skills/corpus",
        "env": {
          "CORPUS_API_TOKEN": "csk_live_or_jwt_token_here",
          "CORPUS_API_BASE_URL": "https://corpusai.app"
        }
      }
    }
  }
}

Commands

Run all commands through:

python3 {baseDir}/scripts/corpus_api.py <command> [options]

Available commands:

• profile
• list-content --limit 20 --cursor <cursor>
• search --query "<text>" --limit 8
• content --user-content-id <id>
• save-url --url <url> [--user-note "<note>"]
• create-reminder --title "<title>" --description "<desc>" --scheduled-date-utc "2026-02-18T16:00:00Z" [--user-content-id <id>]

Recommended workflow for implementation tasks

When a user asks for "find items in Corpus and implement":

1. Use search with a focused query.
2. Use content for top hits to collect concrete steps.
3. Produce an implementation plan with explicit file changes.
4. Apply code changes in the current working repository after user confirmation.

Safety rules

• Never print or log CORPUS_API_TOKEN.
• Prefer read operations before write operations.
• Before write operations (save-url, create-reminder), confirm user intent if the instruction is ambiguous.