The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the existing codebase and conversation history to synthesize technical specifications. If these sources contain malicious instructions, they could influence the generated output or the parameters used in subsequent GitHub CLI commands.
Ingestion points: Reads the current codebase and the conversation context (SKILL.md).
Boundary markers: None. The skill is specifically instructed to 'synthesise what you already know' and 'Do NOT interview the user', which bypasses human verification of the ingested data.
Capability inventory: Execution of GitHub CLI (gh) commands to create and link issues; file system read access.
Sanitization: No explicit sanitization or escaping of the ingested text is performed before it is interpolated into the GitHub issue bodies.
[COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh) to automate the publication of specifications. It executes commands such as gh issue create with various flags (--parent, --repo). This is a standard administrative capability for development agents and targets a well-known service (GitHub).

to-spec