Expanso pii-redact: Claude skill | implexa
back to searchExpanso pii-redact Redact personally identifiable information from text by replacing sensitive data with placeholders using Expanso Edge pipelines.
Create multi-step agents using this skill at Implexa
Implexa builds skills like this into agents that run on a schedule inside your own Claude or Codex, as you, free.
Build an agent with this skill
SKILL.md # pii-redact
"Redact PII from text, replacing sensitive data with placeholders"
## Requirements
- Expanso Edge installed (`expanso-edge` binary in PATH)
- Install via: `clawhub install expanso-edge`
## Usage
### CLI Pipeline
```bash
# Run standalone
echo '<input>' | expanso-edge run pipeline-cli.yaml
```
### MCP Pipeline
```bash
# Start as MCP server
expanso-edge run pipeline-mcp.yaml
```
### Deploy to Expanso Cloud
```bash
expanso-cli job deploy https://skills.expanso.io/pii-redact/pipeline-cli.yaml
```
## Files
| File | Purpose |
|------|---------|
| `skill.yaml` | Skill metadata (inputs, outputs, credentials) |
| `pipeline-cli.yaml` | Standalone CLI pipeline |
| `pipeline-mcp.yaml` | MCP server pipeline |
related skills semantically similar in the cross-vendor index
Create multi-step agents using this skill at Implexa
Implexa builds skills like this into agents that run on a schedule inside your own Claude or Codex, as you, free.
Build an agent with this skill don't have the plugin yet? install it then click "run inline in claude" again.
+ multiple deployment patterns (cli, mcp, cloud) reduce friction for different integration contexts
+ explicit file manifest (skill.yaml, pipeline-cli.yaml, pipeline-mcp.yaml) clarifies artifact structure weaknesses
~ no trigger phrases specified; skill claims pii redaction but never shows what data types or patterns it actually detects (ssn, credit card, email, etc.) ~ procedure section is installation + deployment boilerplate, not actionable steps for pii detection tuning, false-positive handling, or output verification ~ no edge case coverage: silent failures, partial redaction, unicode/encoding issues, rate limits, pipeline timeout behavior all absent ~ no outcome signal defined; unclear how to know redaction succeeded or failed in production