The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze content from the user's codebase and project configuration files, which presents a surface for indirect prompt injection.
Ingestion points: The agent reads arbitrary project files, searches the codebase, and parses the openspec/config.yaml or config.yml file to ground its thinking in project context.
Boundary markers: The instructions explicitly forbid the agent from writing application code or implementing features. It also instructs the agent to treat project rules as constraints rather than content to be reproduced in the conversation or artifacts.
Capability inventory: The skill utilizes the openspec CLI tool via Bash to read project status, list active changes, and write specific design or documentation artifacts like proposals and specs.
Sanitization: There is no evidence of explicit sanitization, filtering, or escaping of the ingested codebase content before it is processed by the agent's reasoning engine.

openspec-explore