The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes a load-time shell command to resolve project-specific configuration keys such as 'rigor', 'workflow', and 'cert_tier'. This execution targets a local helper script (yaml-helper.sh) within the project's directory structure. This is a standard and benign practice for context-aware development tools and is appropriately restricted by the skill's tool permissions.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from local source code comments and bug reports located in production/qa/bugs/*.md. While this ingestion point theoretically allows for embedded instructions to be processed by the agent, the skill uses structured extraction (Grep/Read) to populate a checklist, which presents a minimal risk typical for such tools.\n
Ingestion points: Project source code files (src/, Source/, Assets/) and bug records in production/qa/bugs/.\n
Boundary markers: No specific delimiters or instructions are used to ignore embedded instructions in the ingested files.\n
Capability inventory: The skill has access to Write for saving checklists and Bash for configuration resolution.\n
Sanitization: No explicit validation or filtering is performed on the content extracted from bug reports or code comments.\n- [COMMAND_EXECUTION]: The skill is configured to use the Bash tool to execute a configuration resolution script. The execution is strictly scoped in the frontmatter to a specific file path and a specific subcommand ('resolve_config'), ensuring that the agent cannot execute arbitrary commands.

release-checklist