The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies heavily on external CLI tools to perform its tasks. It uses gh issue create to manage GitHub issues and codex exec to execute code and prompts autonomously. The codex exec command is used with the --sandbox workspace-write flag, allowing it to modify the local filesystem based on model output.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8). It ingests untrusted user data (e.g., task 'braindumps' or feature requests) and interpolates them directly into templates for GitHub issues and prompts for sub-agents.
Ingestion points: User tasks are processed in SKILL.md (Gate 0 and Gate 2) and references/issue-standard.md.
Boundary markers: The skill uses a markdown template for issues but lacks explicit instructions to the implementer model to ignore embedded malicious instructions in the task description.
Capability inventory: The agent can write to the filesystem via codex, create GitHub issues via gh, and spawn sub-agents.
Sanitization: No sanitization or validation of the user's input is performed before it is passed to downstream tools or models.
[EXTERNAL_DOWNLOADS]: The documentation references an installation command using npx to fetch the skill from the author's own repository (thatjuan/agent-skills). This is standard behavior for the skill's distribution model.

ship