The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing various commands via the roborev CLI to perform its primary functions, including roborev init, roborev install-hook, roborev review, and roborev daemon. These commands modify the local environment by installing git hooks and running background processes as part of the intended code review workflow.
[INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface where it processes 'findings' generated by external AI agents (Claude, Copilot, Codex) via roborev show.
Ingestion points: Findings output from roborev show and project-specific .roborev.toml guidelines.
Boundary markers: None explicitly defined in the instructions to delimit finding descriptions from agent instructions.
Capability inventory: The agent can modify the codebase (Fix action), perform further investigations (Discuss), and execute CLI commands.
Sanitization: No specific sanitization or validation of the external finding content is mentioned. While this creates a theoretical surface for indirect injection, it is inherent to the tool's purpose of acting on external review data.

roborev