The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local shell script, scripts/trigger-eval/run.sh, during its workflow. Running scripts directly from a repository can lead to unauthorized code execution if the repository has been maliciously modified.\n- [COMMAND_EXECUTION]: The agent is directed to use git diff commands to retrieve and analyze code changes, which requires access to the system shell.\n- [COMMAND_EXECUTION]: During 'Adopt mode', the skill modifies project-level configurations and installs git hooks. Git hooks are scripts that run automatically during git operations, and their installation can be used as a mechanism for persistent code execution.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from git diffs and pull requests.\n
Ingestion points: Diff outputs from git, source code files, and pull request descriptions.\n
Boundary markers: No explicit delimiters or instructions are used to distinguish between code content and instructions for the agent.\n
Capability inventory: The agent can execute shell commands (git, scripts/trigger-eval/run.sh) and write to files (for configuration and hook installation).\n
Sanitization: No validation or sanitization of the ingested code content is mentioned or performed.

atelier-review-me