The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes the dynamic context injection feature (!command) in its YAML frontmatter to gather project state information. It executes shell commands (echo, cat, ls, head) to retrieve data from production/stage.txt, production/sprints/, and production/session-state/active.md. This is a legitimate use of the feature to provide the AI agent with real-time environment context without incorporating untrusted user input or performing network exfiltration.
[COMMAND_EXECUTION]: Shell commands are executed via the dynamic context injection mechanism to inspect the project structure. These operations are limited to read-only actions on project-specific files and do not involve privilege escalation or sensitive system file access.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect injection as it reads and processes data from external project files like workflow-catalog.yaml, active.md, and sprint-status.yaml. These ingestion points could contain malicious instructions if the project files are compromised. However, the skill primarily uses this data for structural logic (mapping phases) or displaying status snippets. Capability inventory: The skill utilizes Read, Glob, and Grep tools to interact with the file system. Boundary markers: The skill does not implement explicit boundary markers or sanitization when interpolating file content into its output, which is a standard risk for project-tracking skills.

help