OpenParable
Skip to content
← Back to skills

Summarize Document

Documents
summarize-document

Summarize Document is an Agent Skill for documents published in the superlinked/sie repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Summarize a long PDF, scan, office file, text file, or markdown file through the connected Superlinked MCP edge instead of reading the whole source into model context. Use when the user asks for a summary, overview, digest, or "what does this document say" about a large file.

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/superlinked/sie/blob/HEAD/packages/sie_mcp/plugin/claude-code/summarize-document/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/summarize-document, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Summarize a Document with Superlinked MCP

Use the connected Superlinked MCP server to summarize the source on the SIE cluster. This mirrors the summarize-document flow from the sie_tools plugin in PR #1336, but uses the MCP summarize_document tool instead of the gateway-backed ~/.sie/bin/sie wrapper.

Steps

  1. Do not open, view, attach, or read the whole source document into context.

  2. Call the Superlinked MCP summarize_document tool:

    • For PDF, scan, image-of-text, DOCX, PPTX, XLSX, or HTML inputs, read raw bytes only, base64-encode them, and pass document_base64 plus filename.
    • For text or markdown inputs, read raw bytes only, base64-encode them, and pass content_base64.
    • Leave engine as auto unless the user explicitly asks to force a conversion engine.

  3. Write the returned summary to processed/<source-stem>.summary.md. Create processed/ if it does not exist.

  4. Your next message must begin with a short receipt:

    Summarized <source> -> processed/<source-stem>.summary.md (<summary_chars> chars)

    If metadata includes token_savings_estimate, include it as an estimate.

  5. Present the summary from the artifact. If the user asks for exact quotes or detailed follow-up, use parse-document and inspect only relevant sections of the parsed markdown.

Errors

If the MCP call fails, relay the tool error and do not retry blindly. For auth or connector errors, tell the user to reinstall the generated MCP plugin pack or rerun the generated claude mcp add ... --header 'Authorization: Bearer ...' command from INSTALL.md.