The Agent Skills Directory

[SAFE]: The skill establishes a specialized persona for architectural discussion and requirement gathering. It contains explicit instructions that forbid the agent from performing implementation tasks or writing application code, effectively limiting its scope to documentation and analysis within the OpenSpec framework.
[COMMAND_EXECUTION]: The skill instructions include specific bash commands for the openspec CLI to query project status and manage specification artifacts. These operations are consistent with the skill's purpose and are constrained by the environment's tool permissions.
[INDIRECT_PROMPT_INJECTION]: The skill processes project data which serves as a potential ingestion surface for context. 1. Ingestion points: reads config.yaml and existing spec files from the project root. 2. Boundary markers: no explicit markers or boundary warnings are provided in the instructions. 3. Capability inventory: openspec CLI bash commands used for metadata management. 4. Sanitization: no specific sanitization logic is defined for the external project content. The instructions mitigate risk by directing the agent to treat this ingested data as logical architectural constraints rather than content to replicate or execute.

openspec-explore