The Agent Skills Directory

[COMMAND_EXECUTION]: The script assets/domain-drain.py executes external command-line utilities including pdftotext, folder-curator, and hindsight using subprocess.run. These calls use resolved file paths and arguments provided during the triage process.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from inbound artifacts (such as PDFs and Markdown files) to extract text previews and query a memory bank. This external data is used to inform routing decisions and enrich file metadata.
Ingestion points: Inbound files are read in assets/domain-drain.py via the content_preview function.
Boundary markers: The skill bundles extracted data into a JSON decision packet for agent review.
Capability inventory: The skill performs file moves (shutil.move), file writes (write_frontmatter), and external process execution (subprocess.run).
Sanitization: The script strips ANSI escape sequences from tool outputs and uses structured YAML dumping for metadata updates.

domain-triage