The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves analyzing and operating on data from external repositories, which presents a surface for indirect prompt injection attacks.
Ingestion points: The skill reads repository-provided files such as README.md, CHANGELOG.md, package.json, and existing sprint or release documentation during audit and operation modes.
Boundary markers: The workflow relies on the AGENTS.md routing file to guide agent behavior, but it does not implement specific delimiters or 'ignore' instructions for untrusted data processed at runtime.
Capability inventory: The agent has the capability to write to the local filesystem (patching documentation) and execute repository-specific shell commands identified via placeholders like <validation-command>.
Sanitization: There is no explicit sanitization or validation logic described for the content extracted from the analyzed repository before it is used to influence agent decisions or populate templates.

autoscrum