The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it integrates data from an external database into the agent's decision-making process.
Ingestion points: Project state data is retrieved from a database using the get-project command in the project-sync skill (referenced in SKILL.md).
Boundary markers: The instructions do not specify any delimiters or safety markers to differentiate between trusted instructions and data retrieved from the database.
Capability inventory: The skill has the capability to update project state, advance lifecycle stages, and route the agent to downstream skills like reversible-exposure-control or evidence-analysis based on the ingested content.
Sanitization: There is no logic provided to sanitize, validate, or escape the content fetched from the external database before it is used to influence agent behavior.
[DATA_EXFILTRATION]: The skill's protocol for project synchronization involves passing an access-token as a prompt argument (e.g., /featbit-release-decision <project-id> <access-token>). While this is a functional requirement for the project-sync tool, it results in the sensitive token being stored in the user's prompt history. Furthermore, the skill allows the configuration of SYNC_API_URL, which, if modified from its default localhost value, could be used to route synchronization data and tokens to an external endpoint.

featbit-release-decision