The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingest data and instructions from external CLI tool outputs to guide the agent's behavior and content generation.
Ingestion points: The agent is instructed to parse JSON output from openspec instructions, openspec status, openspec context, and openspec schemas (located in SKILL.md).
Boundary markers: There are no explicit boundary markers or instructions to disregard potentially malicious instructions embedded within the CLI tool's JSON responses.
Capability inventory: The skill possesses the capability to execute openspec CLI commands via Bash and perform file-write operations to the local filesystem.
Sanitization: There is no evidence of sanitization or validation of the CLI tool's output before it is used to direct the agent's next steps or fill templates.
[DYNAMIC_EXECUTION]: The skill allows for the execution of commands or skills determined at runtime based on tool output.
Evidence: The instructions state: "If the instruction field delegates creation to a specific skill or command, invoke it to produce the artifact instead of writing the file yourself." This creates a dynamic execution path where the agent executes commands specified in the tool's output data, although the scope is limited by the allowed-tools frontmatter to the openspec command namespace.

openspec-propose