Skip to content
Back to skills

Pp Xray

ASecurity

Parallel Process X-Ray - line-by-line reading of an external module/repo with evidence, risks and calls for the next investigation

  • 2 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 24, 2026
ai-agentspythongobashrailsgit

Security analysis

A100/100

Scanned September 24, 2026

npx -y skills add rusharlabs/house-party-protocol --skill pp-xray --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Pp Xray?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Pp Xray
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/rusharlabs-pp-xray/badge)](https://www.skillsdirectory.com/skills/rusharlabs-pp-xray)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: pp-xray
description: "Parallel Process X-Ray - line-by-line reading of an external module/repo with evidence, risks and calls for the next investigation"
type: skill
---

> **Auto-Trigger:** When the user asks for a technical x-ray, a line-by-line audit, a deep read of an external module or an explanation backed by file evidence.
> **Keywords:** "pp-xray", "x-ray", "line by line", "deep audit", "external repo", "read everything", "end-to-end map"
> **Priority:** HIGH
> **Tools:** Bash, Read, Grep, Glob

# pp-xray - x-ray with evidence

## Goal

Produce a traceable technical reading of a delimited target. Unlike `pp-discovery`, this skill goes into the content of the selected files and records findings with path, line and impact.

Use it after a discovery, or when the user has already provided a scope small enough for a deep read.

## When NOT to Activate

- The target is still too broad for a complete read; use `pp-discovery`.
- The job is to merge reports/agents that have already run; use `pp-consolidate`.
- The question requires current external research; use a research flow with sources.
- The user asked for an immediate implementation/fix; use the appropriate feature/dev skill.

## Process

1. Fix the scope and the completeness criterion.
2. Capture the state:
   - `git status --short --branch`
   - the exact list of files in the target.
3. Read the files in dependency order: manifests/configs, entrypoints, core libs, tests/docs.
4. For each finding, record:
   - file and line;
   - observed fact;
   - impact;
   - uncertainty or pending check.
5. Separate fact from recommendation.

## Expected Output

```md
# PP X-Ray - <target>

## Scope
- included:
- excluded:

## Files Read
| file | lines | role |

## Findings
| severity | file:line | fact | impact |

## Flows
1. input -> processing -> output

## Gaps
| gap | why it blocks certainty | how to verify |

## Recommendations
| priority | action | reason |
```

## Guardrails

- Do not claim "the whole repo" if the scope read was partial.
- Do not edit files during the x-ray, unless explicitly requested.
- Do not replace a live audit with old docs.
- If the target is too large, go back to `pp-discovery` and propose waves.

## Contract

**INPUT:** a delimited target and a completeness criterion.

**OUTPUT:** findings with severity, file:line, fact, impact and gaps.

**EXIT CODES:**

| Exit | Meaning |
|---|---|
| 0 | reading complete within the declared scope |
| 1 | warning: gap declared without invalidating the findings |
| 2 | block: scope too broad or evidence missing |
| 3 | error reading the target |

**STATE IT TOUCHES:**

| Path | Action |
|---|---|
| delimited target | read |
| destination chosen by the operator | write the report |

## Executed examples

```console
$ python -c "print('files_read=4')"
files_read=4
```
<!-- executed: 2026-09-22 · exit=0 -->

```console
$ python -c "print('findings=2 gaps=1')"
findings=2 gaps=1
```
<!-- executed: 2026-09-22 · exit=0 -->

```console
$ python -c "import sys; print('block: missing evidence'); sys.exit(2)"
block: missing evidence
```
<!-- executed: 2026-09-22 · exit=2 -->

## Proof

```bash
python -c "print('files_read=4')"
```

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.