The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands via the openspec CLI, including openspec new, openspec status, and openspec instructions. This allows the agent to interact with the system environment and manage local project files.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and acts upon data retrieved from external CLI command outputs (JSON from openspec instructions), which includes project context, rules, and specific artifact guidance. If these source materials are maliciously modified, they could attempt to influence the agent's output or behavior.
Ingestion points: The skill reads JSON output from openspec status --json and openspec instructions --json commands in SKILL.md.
Boundary markers: The skill explicitly instructs the agent to treat the context and rules fields as internal constraints and to not include them in the generated output files.
Capability inventory: The skill utilizes bash command execution for CLI interactions and file writing capabilities to create artifacts like proposal.md and design.md.
Sanitization: No explicit validation or sanitization is performed on the data parsed from the CLI JSON output before it is used to guide file generation.

openspec-propose