The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently executes the openspec command-line tool to initialize changes, check status, and retrieve instructions. This is a primary function of the skill.
[COMMAND_EXECUTION]: The skill takes user input to derive a project name that is subsequently used as an argument in a shell command (openspec new change "<name>"). Although the skill instructs the agent to convert the input to kebab-case, the use of user-controlled data in a shell environment presents a potential command injection risk.
[PROMPT_INJECTION]: The skill processes structured data from an external CLI tool to generate content, which introduces a surface for indirect prompt injection.
Ingestion points: Data retrieved from the openspec instructions command in SKILL.md (Step 4a).
Boundary markers: The instructions explicitly command the agent not to copy context and rules fields into the output artifacts, establishing a logical boundary for processed data.
Capability inventory: The agent can execute shell commands via the openspec CLI and write files to the local filesystem.
Sanitization: No specific technical sanitization or validation is applied to the data retrieved from the CLI before it is used to generate file content.

openspec-propose