Analysis

• Outcome: Review
• Latest audit: Jul 27, 2026 · 8:45 AM UTC
• Version: 1.0.2

Overview

This is a disclosed local self-audit skill that inspects an agent setup for exposure, drift, cost, and reliability issues, with sensitive behavior bounded by local storage and redaction rules.

Install this only if you want an agent to inspect its own local setup, including sensitive configuration, credential locations, sessions, schedules, and local notes. Review the trigger scope and keep the default proposal-only autofix unless you are comfortable with reversible local hygiene changes being applied and logged automatically.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (6)

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The top-level description contains very broad trigger phrases such as 'check my system', 'run a health check', and 'when something feels off', which can cause the skill to activate in many ordinary conversations. Because this skill is powerful and reads broad local configuration and memory paths, over-triggering increases the chance of unnecessary access to sensitive local files and confusing or unsafe self-audits.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The 'When To Use' section expands activation to loosely defined conditions like 'something feels off' or 'deciding what to fix first', which are subjective and overlap with many normal support tasks. In an agent-routing context, this ambiguity can lead to unintended invocation of a high-privilege diagnostic skill that inspects secrets, permissions, sessions, and other sensitive local state.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill directs the agent to write back persistent spend records in the same turn after any review, without requiring explicit user confirmation before modifying memory. In agent environments, automatic writes to durable state can silently overwrite prior entries, create inaccurate audit history, or let a loosely triggered skill mutate records when the user only asked for analysis.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: This section explicitly states that re-checking the current month overwrites its existing row and never adds a second entry, but it still lacks an upfront safety control such as confirmation, versioning, or append-only history. That makes accidental destruction or silent revision of prior financial records more likely, especially because the skill earlier encourages same-turn updates as part of routine analysis.

Autonomous Decision Making

Medium

• Category: Excessive Agency

Content

## The Reversibility Test

A fix may be applied without asking only if all four hold. Three out of four is a proposal.

1. **Inverse exists and is recorded** — the old value, the old mode, or a copy of the file, written down *before* the change (SKILL.md Rule 5).
2. **Blast radius is one thing** — one file, one setting, one process. Anything that changes behavior for other people or other machines is out.

• Confidence: 92% confidence
• Finding: without asking

Credential Access

High

• Category: Privilege Escalation

Content

name: Analysis
slug: analysis
version: 1.0.2
description: Audits an AI agent's own setup — workspace, config, memory, skills, jobs, integrations — and reports what is broken, exposed, or wasteful. Use when asked to check the system, run a health check, or diagnose the setup, or when something feels off or the agent got slow or expensive; when a token, key, or .env may be exposed in a file, config, or git history; when permissions or auto-approve rules look too broad; when a scheduled job stops firing, runs twice, or fails silently; when sessions or subagents pile up or loop; when memory files bloat, go stale, contradict, or fall out of their index; when skills collide, never activate, or point at missing files; when an integration returns 401 or 429 or goes quiet; when token spend or context size jumps; and when the same finding keeps coming back. Not for vetting third-party skill code (`skill-audit`), workspace persona and proactivity tuning (`openclaw-workspace`), application monitoring (`monitoring`), or statistical analysis of a dataset.
homepage: https://clawic.com/skills/analysis
changelog: "Clearer disclosure of what is stored and where"
metadata:

• Confidence: 78% confidence
• Finding: .env

VirusTotal

64/64 vendors flagged this skill as clean.

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical

• Code: suspicious.exposed_secret_literal
• Location: memory-template.md:61