The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs shell commands to automate repository discovery and file management tasks. Specifically, it uses bash to invoke git and gh for context gathering, and executes find to locate supporting markdown files within the ~/.claude directory.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted external data (project plans and branch diffs) and possesses significant capabilities like file modification and agent orchestration.
Ingestion points: Untrusted content enters the context during Phase 0 from plan files and git diffs.
Boundary markers: No explicit delimiters are used to isolate the untrusted data from the agent's core instructions.
Capability inventory: The skill is granted Bash, Write, Edit, and Agent tools.
Sanitization: The skill does not perform validation or sanitization of the input data before processing.
[SAFE]: The skill implements a defensive backup mechanism, creating timestamped restore points of files before they are edited.

autoplan