The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches information from an external API at https://chat.keri.host/api/chat to provide spec-grounded answers. This is part of the core functionality described in the documentation.
[COMMAND_EXECUTION]: Provides documentation for a curl fallback command, which allows users to interact with the API directly using shell utilities like curl, grep, sed, and jq.
[DATA_EXFILTRATION]: Transmits user-provided messages and base64-encoded file attachments to the vendor's API at keri.host. This represents the intended data flow for the chat and document summarization features.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes data from external sources and user attachments.
Ingestion points: Untrusted data enters the agent context via the attachments array and the Server-Sent Events (SSE) stream from the remote API.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands were found in the skill definition.
Capability inventory: The skill is designed to perform architecture reviews, evaluate spec compliance, and suggest design corrections based on the data it receives.
Sanitization: No sanitization or filtering of the external API response or attachment content is indicated.

chat