ClawHub
Back to skill

Security audit

doc-updates

Security checks across malware telemetry and agentic risk

Overview

This skill is a documentation-maintenance workflow whose repository scanning, doc editing, and optional consolidation steps are disclosed and mostly proportionate to that purpose.

Install this if you want an agent to help keep repository documentation synchronized after code changes. Before running consolidation actions, review every proposed deletion or merge, use the dry-run/skip options when uncertain, and avoid using the broad triggers for unrelated writing work.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The documentation promises exit code 1 when discrepancies are found, but the shown script only prints differences and performs cleanup without any conditional non-zero exit. In automation or CI, this can cause failed sync checks to be treated as passing, allowing stale or incomplete capability documentation to ship unnoticed and weakening documentation quality gates.

Vague Triggers

Medium
Confidence
97% confidence
Finding
The trigger list contains broad, common terms such as "documentation," "writing," and "readme," which can cause this skill to activate in many unrelated contexts. Over-broad invocation increases the chance that the skill will access repository state, inspect files, or steer workflows when the user did not intend to run a documentation-auditing process.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
This module instructs the agent to delete files with `rm <file>` and stage removals via `git add -u`, but it does not require an explicit, user-facing confirmation warning about irreversible data loss before destructive actions occur. In an automation skill that scans for 'stale' or 'redundant' docs using heuristic signals, false positives are plausible, so silent or lightly-confirmed deletion increases the chance of unintended data loss.

Context Leakage

High
Category
Data Exfiltration
Content
9. `doc-updates:accuracy-verified`
10. `doc-updates:preview`

## Step 1: Collect Context (`context-collected`)

- Validate `Skill(sanctum:git-workspace-review)` has been run.
- Use its notes to understand the delta.
Confidence
89% confidence
Finding
The skill explicitly instructs the agent to validate that another review skill has run and to use its notes to understand the delta. That creates cross-skill context propagation, where potentially sensitive workspace review notes, changelog contents, and repository details may be reused more broadly than necessary during a documentation task, increasing the chance of unintended disclosure in generated docs, previews, or summaries.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.