Pro Workflow
Coding-agent workflow with optional Jev hints for correction detection and command risk.
Added to Jevfast
How it uses Jev
When its optional System One classifier is enabled with the Jev provider and a TypeSafe key, the prompt-submit hook asks whether an eligible user prompt corrects or rejects the previous assistant action. Separately, the permission analyst may classify a denied command into a risk class. The first produces a correction hint and the second a report suggestion; neither changes tool permissions, and a failed or slow call returns no classifier answer.
What Jev decides
Detect a user correctionExample answers · not a recorded Jev response · Source ↗
Question 1 · correction
YOUR APP
INSTRUCTION
Is the user correcting or rejecting the assistant's previous action?
STATE
The eligible incoming user prompt string; the hook skips prompts shorter than two words and prompts already matched by its correction regexes.
JEV · NOUL
YesNo
Suggest a denied-command risk classExample answers · not a recorded Jev response · Source ↗
Question 1 · risk
YOUR APP
INSTRUCTION
What is the risk class of this command?
STATE
The command that the permission analyst is reviewing after a denial.
JEV · CHOICE
- read-only
- local-write
- outward
- destructive
App workflow
- Receive a prompt
The prompt-submit hook first checks its correction patterns and classifier configuration.
- Classify when eligible
If the optional classifier is enabled, a TypeSafe key is available, the prompt has at least two words, and no regex already matched, send one correction Noul request.
- Log a hint
At the configured probability threshold, log a correction hint and update correction counts; the user can use /learn to save the lesson.
- Review command risk when requested
The permission analyst may use a Choice result as a risk suggestion, then presents recommendations for user approval.
Why it is interesting
It applies typed judgments to correction tracking and command-risk review inside a coding workflow while leaving learning capture and permission changes under user control.