The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs standard system diagnostic commands (such as df, free, ps, and git) to gather environment and codebase context. These operations are consistent with the skill's stated purpose of ensuring evidence is gathered before any code modifications are proposed.
[PROMPT_INJECTION]: The skill implements a set of "Iron Laws" designed to act as behavioral constraints, instructing the agent to prioritize these rules (e.g., "No Action Without Evidence") over contradictory user requests. This is implemented as a safety guardrail to ensure reliability.
[SAFE]: The included security sub-skill specifically targets and provides remediation for the OWASP Top 10 vulnerabilities. It explicitly warns against hardcoded credentials and recommends secure practices like using environment variables and parameterized queries.
[SAFE]: The skill includes a robust "Indirect Prompt Injection" surface analysis internally, noting that while the agent processes untrusted codebases, the core instructions mandate verification and testing (TDD) phases which serve as mitigation strategies against malicious data in processed files.

kilo-kit-core