The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes an instruction to "Assume the user approves the breakdown and proceed to publish the issues," which directs the agent to bypass a human verification checkpoint for automated write operations. While the skill later mentions iteration until approval, the initial instruction promotes autonomy that can lead to unintended actions.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from external sources and uses it to influence its output.
Ingestion points: Full bodies and comments are fetched from external issue trackers in the context gathering phase (Step 1).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present for the external data.
Capability inventory: The skill has the capability to read from and write (publish) to external issue trackers.
Sanitization: No sanitization or content validation is performed on the ingested issue data before processing.
[DATA_EXFILTRATION]: The skill performs network-based read operations to fetch the content of issues and comments from an external issue tracker based on user-provided references.
[COMMAND_EXECUTION]: The skill utilizes external tools and commands, such as /setup-utarn-skills and automated issue publishing functions, to interact with project management infrastructure.

to-issues-yes