The Agent Skills Directory

[DYNAMIC_EXECUTION]: The test utility scripts/tests/test_generate_weekly_digest.py dynamically loads and executes code from a relative path outside its own skill directory to perform validation.
Evidence: The script attempts to import thesis_store.py from the trader-memory-core skill by constructing a file path using _REPO_ROOT and calling importlib.util.spec_from_file_location followed by exec_module.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from multiple YAML files to generate performance summaries, creating a surface for potential indirect instructions to be included in reports.
Ingestion points: The gather_closed_theses and gather_partial_trims functions in scripts/generate_weekly_digest.py load all files matching th_*.yaml from the state directory.
Boundary markers: The generated Markdown and JSON reports do not use explicit delimiters to isolate untrusted data fields from surrounding report structure.
Capability inventory: The skill is descriptive and limited to local calculation and file writing; it does not possess high-risk capabilities like network access or arbitrary command execution.
Sanitization: The script correctly uses yaml.safe_load() to prevent object injection during the YAML parsing of external data.

weekly-performance-digest