The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from external sources.
Ingestion points: Content is fetched from arbitrary RSS/Atom feed URLs and X/Twitter search results via the xAI API and WebFetch (SKILL.md).
Boundary markers: No specific delimiters or instructions are provided to the agent to isolate the fetched external content from the primary task instructions.
Capability inventory: The skill has the ability to write to the filesystem (memory/), execute shell commands (curl, jq), and send external notifications via a local ./notify tool (SKILL.md).
Sanitization: There is no explicit sanitization or filtering of the ingested text before it is summarized by the LLM.
[COMMAND_EXECUTION]: The skill frequently executes shell commands, including curl, jq, date, and a custom local utility ./secretcurl, to facilitate API communication and data parsing.
[EXTERNAL_DOWNLOADS]: The skill fetches data from external web sources, including the xAI API (https://api.x.ai/v1/responses) and various RSS feeds specified in the user-controlled memory/feeds.yml file.
[DATA_EXFILTRATION]: The skill reads historical logs and user-tracked topics from memory/MEMORY.md and memory/logs/ and transmits summarized information to the user through the ./notify tool.

digest