VexJoy Agent
An agent workspace that routes plain-English requests to specialist agents, skills, and workflows, including an optional Jev-powered route.
An agent workspace that routes plain-English requests to specialist agents, skills, and workflows, including an optional Jev-powered route.
A user invokes /d with a request. Jev classifies it and judges whether the selected route preserves the requested outcome; dispatch proceeds only with an alignment receipt for that proposed intent.
Does the request state an outcome that can be acted on now?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
Does the proposed intent preserve the requested outcome without changing its meaning?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
Does the proposed intent preserve every explicit constraint, limitation, authorization, and requested scope?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
Does the selected route avoid conflicting with the requested outcome?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
Does the proposed intent drop a material deliverable, surface, state, or success condition?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
Does the proposed intent add a material goal, deliverable, or decision absent from the request?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
Is the selected route plainly unable to cover a material part of the request?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
Would execution require the user to resolve an essential ambiguity, contradiction, or reserved choice first?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
Can work proceed because the request is actionable and intent and constraints are preserved?
The state includes the user request, proposed intent, selected agent/skill/pipeline route, attached skills and fan-out agents, plus route validation context. Example selections are illustrative.
The user submits a plain-English task through /d.
Jev classifies the request and checks the proposed route's restatement against the requested outcome.
The router proceeds only when it has an alignment receipt for the exact proposed intent; judgments are recorded without raw request text.
The route is checked against the user's intended outcome before dispatch, and alignment judgments are recorded without retaining raw request text.