The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as its core logic involves reading and interpreting data from external, untrusted project files.
Ingestion points: Per SKILL.md, the agent reads README.md, user-provided specs (e.g., spec.md), and existing docs/*-progress.txt files.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the guidelines for reading these files.
Capability inventory: The skill calls local shell scripts (scripts/init_progress_txt.sh, scripts/archive_plan.sh), creates new files in the docs/ directory, and performs file movements via git mv.
Sanitization: The instructions do not specify any validation or sanitization steps for the data retrieved from repository files.
[COMMAND_EXECUTION]: The skill utilizes local bash scripts to manage planning artifacts.
Evidence: The workflow in SKILL.md explicitly calls for executing scripts/init_progress_txt.sh and scripts/archive_plan.sh.
Logic: These scripts use ripgrep, sed, and git to extract text from files and move documents into an archive folder, which is standard for local file management but relies on parameters derived from user-controlled file content.

arc-implementation-plan-progress