The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands using the git CLI to inspect project status and retrieve file contents (git -C <project-path> diff, git -C <project-path> status, git -C <project-path> show).
[REMOTE_CODE_EXECUTION]: In Step 6, the skill executes lifecycle scripts (typecheck, lint, test) defined in the target project using the detected package manager (bun, pnpm, yarn, or npm). If the target project's configuration is malicious, this results in arbitrary code execution within the agent's environment.
[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection (Category 8). It ingests untrusted data from an external project directory to perform analysis and decision-making.
Ingestion points: The skill reads git diff output and raw file contents from a user-provided project path in SKILL.md (Step 3, 4, and 8).
Boundary markers: No explicit boundary markers or 'ignore' instructions are used when processing the diff or file content.
Capability inventory: The agent has the capability to execute shell commands (via git and package managers) and write to the local filesystem (modifying Lisa templates via the Write tool).
Sanitization: There is no evidence of sanitization or validation of the content retrieved from the target project before it is written back to the Lisa templates.

lisa-learn