Style guide for a grading instrument.
The system is dark, quiet, and editorial: graphite surfaces, calibration marks, candid labels, and modules that make weak instructions feel inspectable.
Color carries the instrument logic.
Graphite
#080908
Page backgroundPanel
#11140f
Cards and modulesPaper
#f4f1e8
Primary textCalibration
#d7ff5f
Primary actions and marksSignal
#8cb8ff
Secondary metadataEditorial
#ff6b6b
Errors and hard warningsCandid hierarchy, no decorative drama.
Grade before you ship.
Clear objective. Bounded agent. Testable result.
Body copy should explain the artifact, the weakness, and the next inspection step without slipping into generic AI optimism.
SPECIMEN RBR-082 / RUBRIC READYButtons, chips, fields, and feedback.
Inputs and status
Atoms become inspection surfaces.
82
Detected artifact type
Verdict
The instruction is usable, but the success gate needs to be explicit before the agent can be trusted in a repeated workflow.
Eval gate module
Objective gate
The output names one measurable success condition.
Drift control
The agent has a fallback path when source evidence is thin.
Format lock
The response follows the requested structure without extra sections.
Free
$0
For quick checks and first rewrites.
Limited audits
Basic score
Top 3 issues
One rewrite
Pro
Preview
For builders who reuse and test instructions.
Full rubric
Advanced rewrites
Eval kit generation
Version comparison
Saved library
Exports
Team
Preview
For shared standards and review workflows.
Team library
Shared rubric library
Admin controls
Private examples
Review reports