evaluate · task skill
Test AI behavior against the task
Evaluate answers, retrieval and actions using representative cases and agreed limits.
Install in your terminal
npx skills add suboss87/fdeops --skill evaluatebring
The AI workflow, representative permitted cases, acceptance criteria and action boundaries.
your agent does
Run the applicable evaluation, inspect failures and check controls for the intended scope.
you get
Evaluation results, failure analysis and a scoped release-evidence verdict.
ask in your agent’s chat
Use FDEOps evaluate on this AI workflow with these representative cases and agreed thresholds. Examine failures and the boundary for permitted actions. [Workflow, cases and criteria]
Replace placeholders with material you may share.
what it produces
5 sections- Evaluation scope
- Cases and thresholds
- Observed results
- Failure analysis
- Scoped verdict
what to check
A technical SHIP verdict is evidence, not production permission. Unresolved critical failures remain visible.
Your agent also reports checks it could not run.
Want fde to coordinate the project?
This task works on its own. Install fde to select instructions across customer work. Installing fde alone does not add each task as a separate menu entry.
npx skills add suboss87/fdeops --skill fdeBring FDEOps into your next customer task.
Use one task skill or let fde help you work across the project. Start with the notes or code you already have.