The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions in SKILL.md guide the agent to execute local Python scripts (triz_suggest.py and graph_manager.py) to generate suggestions and record decisions. Evidence: Shell command examples provided in the 'Protocol' section of SKILL.md for tool usage. The unit test file functions/test_triz_suggest.py utilizes subprocess.run to execute the suggestion script for verification purposes.
[DYNAMIC_EXECUTION]: The functions/triz_suggest.py script dynamically adjusts the Python search path at runtime to import utility functions from a related internal skill. Evidence: Use of sys.path.insert(0, ...) to locate and load graph_manager.py from the nav-graph directory.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided strings describing technical contradictions, which are then used to query a knowledge base. Ingestion points: The contradiction description entered by the user and passed via the --contradiction CLI flag. Boundary markers: The instructions do not define specific delimiters for user input, though the script performs tokenization. Capability inventory: The skill performs local file reads (knowledge graph) and provides structured text output. Sanitization: The triz_suggest.py script employs a restrictive regular expression ([a-z0-9][a-z0-9+\\-]*) to tokenize the input, which serves as a basic filter for non-alphanumeric characters.

nav-triz