The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions direct users to clone a repository from an untrusted GitHub account (sparkfinderoven/r01-hesreallyhim-awesome-claude-code-security).
[REMOTE_CODE_EXECUTION]: The suite facilitates the execution of binaries and scripts located within the downloaded repository (e.g., ~/.claude/skills/security-compliance-suite/bin/owasp-scan), which is an unverified external source.
[COMMAND_EXECUTION]: Several code examples demonstrate the use of subprocess.run (Python) and execSync (Node.js) to trigger shell commands, which are used to interact with the local operating system and network.
[DATA_EXFILTRATION]: The skill provides integration examples that pipe sensitive security scan results (including vulnerabilities and findings) to external Slack webhooks using curl. While this is a documented feature, it establishes a pattern for transferring local security data to remote endpoints.
[DYNAMIC_EXECUTION]: The project uses dynamic code execution patterns to run its security tools, including runtime assembly of shell commands that incorporate file paths and user-provided arguments.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection as it is designed to ingest and analyze untrusted data, including codebase content, architecture documents, and configuration files.
Ingestion points: Vulnerability scanning of the codebase (.), processing of architecture documents (architecture.md), and analysis of dependency manifests.
Boundary markers: The instructions do not define boundary markers or delimiters to prevent the agent from following instructions embedded within the analyzed files.
Capability inventory: The skill possesses capabilities for network operations (curl), shell command execution (subprocess), and file system manipulation.
Sanitization: No sanitization or validation of the content being scanned is mentioned before it is processed by the agent.

sparkfinderoven-security-compliance-suite