The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the openspec command-line interface to execute openspec list --json. This command is used to retrieve the current state and list of active changes within the codebase, which is a legitimate and expected function of the skill.\n- [INDIRECT_PROMPT_INJECTION]: The skill gathers context by reading markdown files such as proposal.md, design.md, and spec.md. These files are external data sources that could theoretically contain instructions intended to influence the agent. The risk is addressed by specific behavioral constraints in the skill's instructions that restrict the agent to conceptual exploration and documentation creation rather than code execution.\n
Ingestion points: Files within the openspec/changes/ directory and specs/ directory (SKILL.md).\n
Boundary markers: Strong instructional guardrails are present, explicitly stating that the agent must not write code or implement features while in explore mode.\n
Capability inventory: The agent can read files, execute the local openspec CLI, and write markdown documentation artifacts.\n
Sanitization: No specific filtering or sanitization of the content from the read files is described.

openspec-explore