The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/run.py dynamically modifies the Python system path (sys.path) at runtime. It searches upward from its current location for a marker file (AGENTS.md) to define a repository root and then inserts this path into the import search order. This pattern, while common in monorepos, constitutes dynamic loading from a computed path which can be exploited if the directory structure is manipulated to redirect imports to malicious modules in the tooling namespace.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted data from multiple files.
Ingestion points: The skill reads data from outline/module_plan.yml, sources/index.jsonl, and sources/provenance.jsonl (found in scripts/run.py).
Boundary markers: No explicit delimiters or instructions to ignore embedded prompts are present in the skill definition.
Capability inventory: The skill possesses file system read and write capabilities (load_yaml, write_jsonl in scripts/run.py).
Sanitization: No evidence of sanitization, validation, or escaping of the ingested source/plan content is provided in the script boundary.

module-source-coverage