The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use local search tools such as ripgrep (rg) and grep to explore the codebase for security-relevant patterns. These operations are restricted to the local repository path provided by the user and include best-practice flags (such as -I) to exclude binary data from analysis.
[PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection because it processes untrusted data from the user's codebase. Malicious instructions embedded in analyzed files could potentially attempt to influence the agent's behavior during the modeling process.
Ingestion points: Files within the local repository or directory path provided by the user.
Boundary markers: Uses structured prompt templates and markdown fencing to delimit analysis.
Capability inventory: Uses file system read access via standard CLI search tools and writes the final output to a markdown file.
Sanitization: The skill includes explicit instructions to redact secrets and credentials encountered during analysis and mandates that claims must be grounded in evidence.
[SAFE]: No malicious patterns such as credential exfiltration, remote code execution from untrusted sources, or obfuscation were detected. The skill follows security best practices by requiring user confirmation for assumptions before finishing the report and explicitly separating runtime behavior from development tooling.

security-threat-model