ClawHub
Back to skill

Security audit

skill-gardener

Security checks for vulnerabilities and agentic risk

Overview

The skill is a disclosed local skill-maintenance workflow with a bounded audit helper, but it can edit user-owned skill files when authorized.

Install this only if you want an agent to help maintain local skills. Keep automatic gardening disabled unless you are comfortable with the agent making scoped edits to user-owned skill files after validation, and review proposed diffs for any workflow that could affect future agent behavior.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description says the skill creates or repairs local skills from verified workflows. The supplied code is strictly a read-only auditing script: it enumerates directories, reads SKILL.md files, safely parses YAML frontmatter, validates structure and field constraints, detects duplicates/symlinks/depth issues, and emits a JSON audit report. There are no write operations to create, update, repair, or save any skill content. This is a materially different primary purpose, so the description does not accurately represent the code.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 20)May include surrounding context.

md
evant evidence and nearby catalog matches. A collection audit may read bounded `SKILL.md` files under the selected skill root; it must not expand into unrelated

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 93)May include surrounding context.

md
evant evidence and nearby catalog matches. A collection audit may read bounded `SKILL.md` files under the selected skill root; it must not expand into unrelated

Insecure deserialization: yaml.load() with an unsafe Loader

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

Untrusted data is passed to an insecure deserializer (pickle, marshal, dill, jsonpickle, joblib, yaml.load without a safe Loader, or torch.load without weights_only). These deserializers reconstruct arbitrary objects and invoke callables during loading, so deserializing attacker-controlled bytes is equivalent to arbitrary code execution.

Content

Scanner excerpt · scripts/audit_skills.py (reported line 277)May include surrounding context.

python
return {}, "skill body is empty"
    try:
        source = "\n".join(lines[1:end]) + "\n"
        fields = yaml.load(source, Loader=FrontmatterLoader) if yaml is not None else fallback_load(source)
    except ((yaml.YAMLError if yaml is not None else FrontmatterError), ValueError, OverflowError, RecursionError) as exc:
        mark = getattr(exc, "problem_mark", None)
        location = f" near line {mark.line + 2}, column {mark.column + 1}" if mark else ""

Scope Creep

Low
Category
Excessive Agency
Confidence
70% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · scripts/audit_skills.py (reported line 394)May include surrounding context.

python
issue(path, "allowed-tools must be a non-empty string or list of non-empty strings when provided")
        if isinstance(fields.get("compatibility"), str) and len(fields["compatibility"]) > 500:
            issue(path, "compatibility exceeds 500 characters")
        # OpenClaw accepts nested metadata; it is not limited to flat strings.
        if "metadata" in fields and fields["metadata"] is not None and not isinstance(fields["metadata"], dict):
            issue(path, "metadata must be a mapping when provided")

Static analysis

No suspicious patterns detected.