The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from various external NYC Open Data (Socrata) APIs. This introduces a vulnerability surface where the agent could potentially be influenced by malicious content embedded in the queried datasets, although the risk is mitigated by the official nature of the data sources.
Ingestion points: External data is fetched via the WebFetch tool from several Socrata endpoints (e.g., data.cityofnewyork.us/resource/64uk-42ks.json, buis-pvji.json, ipu4-2q9a.json, etc.) as defined in SKILL.md and socrata-reference.md.
Boundary markers: The skill uses a structured markdown template in SKILL.md to organize the report, which provides some separation between the fetched data and the agent's core instructions.
Capability inventory: The skill has access to WebFetch, Write, Read, and Bash tools, providing significant capabilities to interact with the network and the local filesystem.
Sanitization: There are no specific instructions within the skill to sanitize, filter, or escape the JSON content retrieved from the external APIs before it is interpolated into the markdown report or presented to the agent context.

nyc-property-report