The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests and processes untrusted product intent, PRDs, and feature specifications to drive the project lifecycle. These inputs are interpolated into the agent's context, creating a surface for indirect prompt injection attacks.
Ingestion points: The skill reads data from docs/PRD.md, docs/feature-breakdown.md, specs/*/spec.md, and user-provided product intent descriptions.
Boundary markers: No explicit boundary markers (e.g., XML tags) or instructions to ignore embedded commands are used when the agent processes these artifacts.
Capability inventory: The skill can perform file system operations (writing specs and plans), execute shell commands via the specify CLI, and orchestrate other skills like /shiplight.
Sanitization: There is no evidence of sanitization or structural validation performed on the ingested content before it is used to influence the agent's planning and implementation steps.
[EXTERNAL_DOWNLOADS]: The skill references and integrates with external tools and repositories.
Trusted Source: References the GitHub Spec Kit documentation and repository located at github.com/github/spec-kit. This targets a trusted organization.
Installation Command: The documentation suggests installing the skill using npx skills add ShiplightAI/quality/agent-skills, which uses the skills package from the official NPM registry.
[COMMAND_EXECUTION]: The skill orchestrates the project workflow by executing external commands and other agent skills.
Tool Delegation: It invokes the specify CLI for Spec Kit mode and the /shiplight skill (using commands like /shiplight cover and /shiplight verify) for testing and verification tasks.

spec-project