The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the vendor-owned @nomadamas/k-skill package from the NPM registry as part of the setup and execution process described in SKILL.md.
[REMOTE_CODE_EXECUTION]: Executes the @nomadamas/k-skill CLI tool using npx to manage skill instructions and helper execution. This is a standard vendor-provided utility for this skill's ecosystem.
[COMMAND_EXECUTION]: Invokes the local Python script scripts/mfds_drug_safety.py to handle the logic for the mandatory safety interview and the API communication flow.
[DATA_EXFILTRATION]: Performs network requests to https://k-skill-proxy.nomadamas.org, a vendor-controlled domain, to retrieve official drug information. This communication is essential for the skill's primary function.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing external API data.
Ingestion point: scripts/mfds_drug_safety.py retrieves data from a remote proxy via the read_json_response function.
Boundary markers: No explicit prompt boundaries or "ignore instructions" markers are defined for the interpolated API data in the instructions.
Capability inventory: The script has capabilities for network access (urllib.request) and command-line output.
Sanitization: The script implements basic sanitization using html.unescape and re.sub(r"<[^>]+>", " ", text) to strip HTML tags from ingested data before it is returned to the agent context.

mfds-drug-safety