The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell commands (bash, jq, bc, sed, tr) and Python one-liners to perform state management, mathematical synchronization score calculations, and message dispatching. It specifically utilizes the GitHub CLI (gh) for issue-based anchoring and the 'maw' tool for cross-agent messaging.
[EXTERNAL_DOWNLOADS]: The skill interacts with GitHub's public API through the gh tool to view and comment on issues. These operations are limited to metadata retrieval and documentation within well-known project management contexts.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface (Category 8) because it processes data from external sources such as GitHub issue comments and incoming synchronization messages.
Ingestion points: Data enters the context via GitHub comments retrieved by gh issue view and protocol messages (SYNC-RESULT, PARTY INVITE) received via the maw hey command.
Boundary markers: The implementation uses specific protocol headers (e.g., SYNC-CHECK, SYNC-RESULT) and structured JSON schemas to delineate messages.
Capability inventory: Capabilities include writing to specified memory directories and sending messages to other oracles.
Sanitization: Data is filtered and parsed using jq and standard string manipulation tools, which limits the potential for executing arbitrary instructions from external content.

work-with