The Agent Skills Directory

[DYNAMIC_EXECUTION]: The workflow utilizes edgar-analyzer generate-code to produce scripts at runtime based on patterns identified in project examples, which are then executed via edgar-analyzer run-extraction to perform data processing.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted external data from SEC EDGAR filings (e.g., DEF 14A HTML). While the process is structured, it represents an attack surface where adversarial content in a filing could attempt to influence the extraction logic or agent behavior.
Ingestion points: SEC EDGAR filing data (HTML/Text) fetched from public SEC servers.
Boundary markers: Not explicitly defined in the high-level documentation; the tool uses specialized adapters ('SCTAdapter') for parsing.
Capability inventory: Network operations to fetch filings, file system writes for reports (CSV/JSON), and execution of generated Python scripts.
Sanitization: Not specified; the skill relies on the parsing logic of the adapters to isolate relevant data.
[COMMAND_EXECUTION]: The skill relies on the edgar-analyzer CLI tool and several documented Python scripts (fetch_apple_def14a.py, create_csv_reports.py) to manage the extraction and reporting lifecycle.

sec-edgar-pipeline