The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill processes external data from project management platforms, including work item bodies and user comments. This content is inherently untrusted and could contain malicious instructions. The skill mitigates this by including explicit instructions to treat all fetched payloads as untrusted and to report embedded directives as observed content rather than executing them.
[Credential Management]: To interact with external services, the skill requires authentication tokens. It appropriately identifies these as environment-based requirements rather than hardcoding sensitive information, which aligns with standard security practices for secret management.
[Local State Persistence]: The skill maintains workflow state by reading and writing planning files to a local directory. This enables the resumption of tasks across sessions without performing unauthorized modifications to the external trackers.

backlog-plan