The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands for file management and execution.
Evidence: Uses ls, mkdir, and mv to manage files within the plans/ directory.
Evidence: Executes python3 ~/Documents/agent-skills/scripts/plan_runner.py report <plan> to generate execution summaries.
[DYNAMIC_EXECUTION]: The skill relies on an external local script to process data and generate content for the plan files.
Evidence: The script ~/Documents/agent-skills/scripts/plan_runner.py is called at runtime. While the path is static, the behavior depends on the integrity of this local script.
[INDIRECT_PROMPT_INJECTION]: The skill reads, parses, and modifies user-provided markdown files, which could contain instructions that influence the agent's parsing logic or archival behavior.
Ingestion points: Reads files from plans/active/ and state files from .plan-state/ (SKILL.md Step 2 and 2.5).
Boundary markers: None identified; the agent parses the text based on markdown headers and specific patterns (e.g., S-code format).
Capability inventory: Bash execution, local script execution via Python, and file writing/moving capabilities.
Sanitization: No explicit sanitization or validation of the plan content is mentioned before it is parsed or before the execution summary is embedded.

plan-archive