The Agent Skills Directory

[Data Exposure & Exfiltration] (SAFE): The skill performs read-only network requests to hn.algolia.com, a public API. No credentials are required or exposed, and no sensitive local data is accessed or transmitted.
[Indirect Prompt Injection] (SAFE): The skill is designed to process untrusted user-generated content (Hacker News comments and stories). This is the primary intended use case.
Ingestion points: SKILL.md (JSON response parsing from the Algolia API).
Boundary markers: Not explicitly defined in the instructions, though the skill provides structured formatting for the output.
Capability inventory: Only includes web_fetch for data retrieval and exec: date +%s for local timestamp calculation; no destructive or persistent capabilities are present.
Sanitization: Not specified, but the agent's presentation format acts as a display layer.
[Command Execution] (SAFE): The skill uses exec: date +%s to generate a current Unix timestamp. This is a standard, low-risk operation used to calculate date filters for the API query.

hn-search