The Agent Skills Directory

[SAFE]: The skill operates entirely on local data stored in the .job-hunter/ directory. No network operations, external requests, or telemetry patterns were identified, ensuring user privacy and data locality.
[COMMAND_EXECUTION]: Deterministic detectors flagged the use of subprocess.run() in the test files (tests/test_harvest_outcomes.py, tests/test_propose_lessons.py). These calls are benign as they are used solely to execute the skill's own scripts during automated testing.
[PROMPT_INJECTION]: The skill ingests user-controlled markdown data (such as rejection reasons in OUTCOMES.md). Security is maintained through deterministic keyword-based analysis and a templated suggestion system that prevents raw user input from being interpreted as agent instructions.
[SAFE]: The skill implements significant safety guards, including a 'cold-start' threshold (requiring 5+ outcomes) and an explicit opt-in requirement for all permanent changes to the user's LESSONS.md file.

outcome-learning