The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes Git commands (git remote, git checkout, git pull) to manage the repository and updates a local state file using the marsh CLI or jq for atomic writes to ~/.config/marshroom/state.json.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) when processing data from the Marshroom state file. Ingestion points: The issueBody field is read from ~/.config/marshroom/state.json and displayed to the agent. Boundary markers: None; the content is injected into the agent's context without delimiters or instructions to ignore embedded commands. Capability inventory: The skill has the ability to manipulate Git branches, pull code from remotes, and modify local configuration files. Sanitization: No validation or escaping is performed on the issueBody content before it is processed by the agent.

start-issue