Openclaw Skills for redact-pii masks sensitive text in-line with ZeroGPU, replacing detected PII with bracketed labels before you share or log it.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install redact-pii
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install redact-pii using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is redact-pii?
redact-pii is an Openclaw Skills utility for in-line PII sanitization. It sends raw text to ZeroGPU and returns the same passage with detected sensitive spans replaced by labels such as [PERSON] and [EMAIL].
It is built for fast, low-friction redaction in agent workflows: the text is passed verbatim through a safe heredoc wrapper, the CLI performs the masking, and the output preserves the original structure while hiding private identifiers. If you need to recover entities instead of masking them, use the companion extract-pii workflow.
redact-pii Use Cases
- Redacting customer support messages before sharing them in tickets or chat.
- Scrubbing logs, telemetry, or error reports to remove personal data.
- Anonymizing prompts, transcripts, or notes before publishing or indexing.
- Sanitizing copy pasted from emails, forms, or documents before escalation.
- Preparing sensitive snippets for Openclaw Skills agent workflows that require safe text.
How redact-pii Works
- Receive the user’s raw text as
$ARGUMENTS. - Wrap the content in a single-quoted heredoc so newlines, quotes, and shell metacharacters survive unchanged.
- Pass the resulting string to
zerogpu redact_pii "$ZGPU_TEXT"through the allowed Bash tool. - ZeroGPU detects PII spans and replaces each one with a bracketed placeholder such as
[PERSON]or[EMAIL]. - Return the original text shape with sensitive values masked; if ZeroGPU emits a literal
💰 ZeroGPU savingsline, append it unchanged as the last line only.
redact-pii Setup
- Make sure the Openclaw Skills runtime can invoke the ZeroGPU CLI from Bash.
- No extra model files or local configuration are required for this skill; it runs entirely through
zerogpu redact_pii. - Feed raw text directly via the heredoc pattern below so all special characters are preserved.
ZGPU_TEXT=$(cat <<'ZGPU_END_OF_INPUT'
Your raw text with names, emails, or other PII here.
ZGPU_END_OF_INPUT
)
zerogpu redact_pii "$ZGPU_TEXT"
- Validate that the response contains placeholder tokens like
[LABEL]and no exposed sensitive values.
redact-pii Data Schema & Taxonomy
| Layer | Data | Purpose |
|---|---|---|
| Input | $ARGUMENTS |
Raw user text passed verbatim into the redaction flow. |
| Transport variable | ZGPU_TEXT |
Holds the heredoc-wrapped payload so multiline content and shell metacharacters are preserved. |
| Processing engine | zerogpu redact_pii |
Detects PII spans and converts them into placeholder labels. |
| Output | Redacted text | The same passage with sensitive spans replaced by tokens like [PERSON] and [EMAIL]. |
| Files generated | None | The skill is in-memory only; it does not write artifacts to disk. |
| Metadata taxonomy | Placeholder labels | ZeroGPU decides the entity classes; typical examples include person names, emails, and other PII categories. |
Notes
- No persistent config files, manifests, or cache artifacts are created by the skill.
- The output preserves source formatting while changing only the detected sensitive spans.
redact-pii Advanced Features
- Safe multiline handling via a single-quoted heredoc, which preserves quotes, newlines, and shell metacharacters.
- In-line redaction that keeps document structure intact while hiding only the detected PII spans.
- Zero-disk workflow: no files are created, so the skill is easy to use in transient agent sessions.
- Placeholder-based output that is easy to re-ingest, diff, or paste into logs and tickets.
- Companion workflow compatibility with
extract-piiwhen you need entity extraction instead of masking. - Conditional savings footer passthrough for the literal
💰 ZeroGPU savingsline when the backend emits it.
SKILL.md
Loading
Related Openclaw Skills
Social Arbitrage
Openclaw Skills turns real-world social and consumer shifts into evidence-backed U.S. equity research with a skeptical, Chris Camillo-inspired workflow.
World Builder
A meta-skill that turns a world idea or character set into a coherent, 12-dimensional world specification.
CertainLogic Stack
CertainLogic Stack is an Openclaw Skills AI stack that combines verified knowledge, low-cost routing, caching, and audit trails to cut spend and stop hallucinations.
Aesthetic Builder
Aesthetic Builder converts a concept or character set into a structured 9D aesthetic system with dual-mode generation and extraction.
Classify Structured
Schema-driven multi-axis text classification that turns arbitrary text into structured JSON labels with Openclaw Skills.
Extract Entities
Openclaw Skills turns ZeroGPU into a fast, custom-label NER workflow for extracting entities from any passage of text.
Featured*
