The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements an indirect prompt injection surface through its automated hooks.
Ingestion points: The PreToolUse hook executes cat task_plan.md 2>/dev/null | head -30 || true whenever the Write, Edit, or Bash tools are about to be used. This reads the task plan directly into the agent's active context.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the command output displayed to the agent.
Capability inventory: The skill permits use of Bash, Write, Edit, WebFetch, and WebSearch, which provides both the means to fetch untrusted data and the ability to execute system commands.
Sanitization: No sanitization or validation of the file content is performed before it is injected into the session context via the hook.
[COMMAND_EXECUTION]: The skill uses shell commands within multiple lifecycle hooks to automate the planning workflow.
The PreToolUse hook executes a shell pipeline (cat, head) to display the current plan.
The Stop hook executes a shell script located at ${CLAUDE_PLUGIN_ROOT}/scripts/check-complete.sh to verify task completion.

planning-with-files