Incomplete referenced artifact analysis
High
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md Full sources for every rule: `references/sources.md`.
Security audit
video-script
Security checks for vulnerabilities and agentic risk
Overview
ClawScan rates this skill suspicious with medium confidence, mainly because its documented install command uses an unpinned package and a recommended judging workflow may send scripts and briefs to external AI providers; one referenced file was only partially inspected.
Consider avoiding installation until you can verify and pin the installer and skill version. If using the judging workflow, check whether the selected AI provider is approved for your material, especially if it is confidential; the scan was partial, so some evidence remains incomplete.
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
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
- Rogue AgentSelf-Modification, Session Persistence
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)
Incomplete referenced artifact analysis
High
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md Story Spine, Harmon's circle and Save the Cat are in `references/sources.md` as audits, not molds.
Self-Modification
High
- Category
- Rogue Agent
- Confidence
- 85% confidence
- Finding
Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.
- Content
md ## 1. Keep a test set `references/test-briefs.md` holds six briefs, one per format. Before you edit this skill, write all six with the current version and keep the files as the baseline. "Create evaluations BEFORE writing extensive documentation" ([Anthropic, skill best practices](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/best-practices)).
MCP server referenced without pinned version: 'npx skills'.
Medium
- Category
- MCP Rug Pull
- Confidence
- 90% confidence
- Finding
The installation command executes an unpinned npm package through npx, exposing users to compromised future releases executing with their local permissions. This is a supply-chain weakness in the documented installation path, not an MCP server reference as the detector claims; there is no evidence the current package is malicious.
- Content
The judging workflow sends brief and script content to LLM judges without a privacy warning.
Medium
- Category
- Output Handling
- Confidence
- 88% confidence
- Finding
The workflow recommends judges from other model families and supplies prompts containing briefs and scripts, without checking whether those providers are approved to receive the material. For confidential launch plans or unreleased product claims, following this workflow with an external judge could disclose sensitive information; the risk depends on the chosen deployment and existing authorization.
- Content