The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a local shell script scripts/init-plan.sh to scaffold the project structure. The script is well-authored, using set -euo pipefail and safe string handling for directory and file creation.
[PROMPT_INJECTION]: The skill instructs the agent to read and follow state information from the .plan/ directory, which constitutes an indirect prompt injection surface. 1. Ingestion points: .plan/task_plan.md, .plan/findings.md, and .plan/progress.md are read to maintain context. 2. Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the templates or rules. 3. Capability inventory: The agent has file system access and the ability to execute shell commands. 4. Sanitization: There is no requirement or mechanism provided to sanitize or escape untrusted data (e.g., from research findings) before it is written to these persistent planning files.

planning