The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its data ingestion process. * Ingestion points: The agent is instructed to read external files including task packets (tasks/{{TASK-slug}}.md), feature specifications (specs/{{feature}}/spec.md), and repository command configurations (AGENTS.md). * Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between its own system instructions and potentially malicious commands embedded within the external task data. * Capability inventory: The skill allows the agent to modify the codebase, create new files (references/task-template.md), and execute shell commands defined in the project's AGENTS.md file. * Sanitization: The skill lacks mechanisms for sanitizing or validating the contents of the ingested files before the agent processes them.

write-feature