The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple local shell commands using the openspec CLI (e.g., openspec new, openspec status, openspec instructions) to manage project state and retrieve metadata. These commands are integral to the skill's stated purpose of automating document generation.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface because it processes and follows instructions provided by external CLI tool outputs at runtime.
Ingestion points: Step 4a in SKILL.md involves reading JSON formatted instructions, rules, and context from the openspec instructions command.
Boundary markers: Explicit delimiter boundaries are absent; however, the skill provides logical separation by instructing the agent to use the data as constraints rather than directly copying it into the generated files.
Capability inventory: The skill has the capability to execute subprocesses via the openspec CLI, write files to the local filesystem, and query the user for input using the AskUserQuestion tool.
Sanitization: No specific sanitization or validation of the ingested JSON data is performed before it is used to influence the agent's content generation.

openspec-propose