The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the meegle command-line interface to interact with Feishu Project data. It specifically instructs the agent to run meegle chart get for data retrieval and meegle auth login to handle session expiration. These operations are restricted to specific, hardcoded project keys and chart IDs, reducing the risk of arbitrary command injection.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it fetches and interprets external data from a collaborative project management environment.
Ingestion points: Untrusted data enters the agent context through the JSON output of the meegle chart get command, which includes product names, labels, and metadata as described in SKILL.md and references/1-data-discovery.md.
Boundary markers: The instructions lack explicit boundary markers or "ignore embedded instructions" warnings for the agent when processing the fetched chart content.
Capability inventory: The skill's primary capabilities include generating diagnostic reports, ranking lists, and providing natural language explanations based on the ingested data.
Sanitization: There are no documented procedures for sanitizing or validating the external strings retrieved from the Feishu Project charts before they are presented to the user or processed by the agent.

meegle-rdhs-diagnosis