The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references an external documentation URL (jeffallan.github.io/claude-skills/skills/devops/cli-developer/) which has been identified as a malicious URL by automated security scanners (URLite Blacklist).
[METADATA_POISONING]: Automated file reputation scans have flagged SKILL.md as potentially malicious (FileRepMalware). Additionally, there is a discrepancy between the author listed in the file metadata (Jeffallan) and the assigned developer context (emmraan), indicating a potential supply chain risk or impersonation attempt.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a code generation assistant for CLI tools, ingesting untrusted user specifications for command structures, flags, and logic. This creates a surface for indirect prompt injection where malicious input could influence the generated implementation.
Ingestion points: User-provided CLI requirements and metadata used to populate code templates in SKILL.md and referenced files.
Boundary markers: No explicit delimiters or instructions to ignore embedded prompts are present in the generation logic.
Capability inventory: The skill generates code that utilizes powerful system APIs, including file system access (fs-extra, path) and process execution (execa, subprocess.run), which could be abused if malicious instructions are successfully injected into the generation process.
Sanitization: No evidence of validation, escaping, or filtering of user-provided strings was found in the provided templates.

cli-developer