The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard system commands including git and gh (GitHub CLI) to manage pull requests and environment branches. It also invokes a local binary, fluxctl, located within the .opencode/bin/ directory. These operations are restricted to the local workspace and official project configuration.
[DATA_EXFILTRATION]: The skill reads deployment metadata and branch configurations from .flux/config.json. This data is used solely for determining target branches and staging URLs within the internal workflow.
[PROMPT_INJECTION]: The skill incorporates acceptance criteria from project epics to guide automated testing.
Ingestion points: Acceptance criteria are retrieved from recently completed epics in the project context.
Boundary markers: The instructions do not define specific delimiters to isolate the external epic content from the core validation logic.
Capability inventory: The agent has access to git for repository management, gh for PR creation, and the local fluxctl utility.
Sanitization: No specific sanitization logic is implemented for the epic content; however, the impact is limited to validation reporting and human-reviewed PR creation.

flux-opencode-gate