The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized network operations were detected. The skill is scoped to the openspec CLI tool, which aligns with its metadata and description.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it ingests data from external sources during execution.
Ingestion points: Data is ingested through the output of openspec status and openspec instructions JSON, as well as the contents of files specified in the contextFiles array.
Boundary markers: The skill includes explicit instructions for the agent to treat context and operationGuidance fields as separate prompt-level inputs that cannot override CLI-controlled state or bypass safety gates.
Capability inventory: The agent has the capability to execute specific shell commands via the openspec CLI and perform targeted file modifications (updating task status).
Sanitization: There is no explicit sanitization step for the ingested data, but the usage constraints act as a functional safeguard.

openspec-apply-change