The Agent Skills Directory

[PROMPT_INJECTION] (SAFE): No instructions were found that attempt to bypass AI safety filters or override agent behavior.
[DATA_EXFILTRATION] (SAFE): While the skill inventories sensitive assets (credentials, PII) for documentation, it contains no network calls or mechanisms to transmit this data externally.
[COMMAND_EXECUTION] (SAFE): Verification steps utilize standard, low-risk Unix utilities (grep, jq, test) to validate the structure of locally generated reports.
[REMOTE_CODE_EXECUTION] (SAFE): No remote scripts are downloaded or executed.
[INDIRECT_PROMPT_INJECTION] (SAFE): The skill ingests untrusted codebase data as its primary function. Because the skill's capabilities are limited to local file writing and lacks network access or high-privilege execution, the risk of an attacker influencing the agent via malicious code comments is minimal.

threat-model-generation