OpenParable
Skip to content
← Back to skills

Redact Pii

DevOps & Security
redact-pii

Redact Pii is an Agent Skill for devops & security published in the superlinked/sie repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Redact personal data from a document through the connected Superlinked MCP edge before working with the content. Use when the user asks to redact, anonymize, scrub, de-identify, or remove PII/sensitive data from a document.

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/redact-pii/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/redact-pii, 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.

Redact PII with Superlinked MCP

Use the connected Superlinked MCP server to detect and replace PII on the SIE cluster. This mirrors the redact-pii flow from the sie_tools plugin in PR #1336, but uses the MCP redact_pii tool.

Hard Rules

  • Do not open, view, attach, or read the source document into context before redaction.
  • After redaction, continue the user's task using only the redacted artifact.
  • The MCP tool intentionally does not return a placeholder-to-original map. Do not promise de-redaction from the MCP flow.

Steps

  1. Call the Superlinked MCP redact_pii 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.
    • Pass labels only if the user wants a custom PII/sensitive-data set.
    • Leave engine as auto unless the user explicitly asks to force a conversion engine.

  2. Write the returned redacted_text to processed/<source-stem>.redacted.md. Create processed/ if it does not exist.

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

    Redacted <span_count> spans from <source> -> processed/<source-stem>.redacted.md

    Include the label counts and the note that detection is model-based and best-effort, not a certified scrubber.

  4. Continue the user's actual task using only the redacted artifact.

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.