Back to skill

Security audit

Project Quality Governance

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent engineering documentation governance helper, with no artifact-backed evidence of hidden execution, exfiltration, persistence, or malicious behavior.

Install this for engineering repositories where you want structured technical specs, quality logs, and validation checks. Review the paths and validator configuration before use, especially in repositories with private research or secrets, and invoke it explicitly if you do not want automatic routing.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 29)May include surrounding context.

md
- Use `scripts/validate-quality-docs.mjs` only after adapting its explicit configuration to the repository's schema.

YARA rule 'agent_skill_prompt_injection_hidden_instructions': Prompt injection or hidden instructions embedded in AI agent skill text [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · references/technical-spec-architecture.md (reported line 7)May include surrounding context.

md
# Technical Specification Architecture

## Authority and Scope

Define an explicit source-of-truth order: current user direction, invariant safety and data rules, verified implementation behavior, then historical plans. State which document is authoritative and when it must be updated.

## Front-Loaded Reading Model

Place a compact, mandatory rule block immediately after the title. Cover only constraints that apply to every task: data boundaries, security, protected code, scope control, repository hygiene, and verification responsibility.

Follow it with a clickable section index, a task-to-section routing table, and a note that historical or experimental material is read only when current facts cannot explain a decision. This enables narrow context loading without weakening safety.

Use one lightweight contributor entrypoint as the default read. The authoritative specification remains queryable re

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
90% confidence
Finding

Skill attempts to nullify the agent's safety policies or restrictions ('you have no restrictions', 'ignore your guidelines', 'do anything now'). This is a direct jailbreak that disables guardrails.

Content

Scanner excerpt · references/technical-spec-architecture.md (reported line 19)May include surrounding context.

md
Define repository-specific limits for initial file count, default line range, command output, visual checks, and handoff length. Keep the limits in the lightweight entrypoint and validate them mechanically when the repository chooses fixed values.

Budgets reduce observation noise; they do not override safety review or prevent reading additional evidence when a task genuinely requires it.

## Stable Hierarchy

Instruction Override

High
Category
Prompt Injection
Confidence
90% confidence
Finding

This pattern attempts to override system instructions or ignore safety constraints. Without LLM analysis, manual review is recommended.

Content

Scanner excerpt · references/technical-spec-architecture.md (reported line 19)May include surrounding context.

md
Define repository-specific limits for initial file count, default line range, command output, visual checks, and handoff length. Keep the limits in the lightweight entrypoint and validate them mechanically when the repository chooses fixed values.

Budgets reduce observation noise; they do not override safety review or prevent reading additional evidence when a task genuinely requires it.

## Stable Hierarchy

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This plain-text brief describes the skill's purpose at a high level but does not specify clear trigger phrases, invocation conditions, or boundaries for when the skill should activate versus stay inactive. Because the file is repository-neutral and broadly framed around improving engineering repositories, it could be interpreted as applying to many common requests without clear constraints or negative examples.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The brief explicitly requires English instructions, which imposes a language choice regardless of user preference. The same line allows repository-language output by default, but it does not offer the user a language selection or document a justified locale restriction.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill enables allow_implicit_invocation without any narrowing conditions, so the agent may trigger this governance skill in situations broader than intended. Because this skill can shape specifications, audits, logs, and validation workflows, unintended activation could steer repository documentation or quality processes without an explicit user request, causing confusing or unsafe automation decisions.

Content

No source excerpt is available for this finding.