Use when driving a web page in a browser — clicking, typing, navigating, logged-in or JS-rendered pages. Jev picks each step from the elements observed, under a host allowlist and a step budget.
中文简介正在准备中。当前显示原文说明。
Use for all Blender work. Default code-first execution, local QA and bounded Jev diagnostic advice.
中文简介正在准备中。当前显示原文说明。
安装前可以查看交给 AI 代理的指令内容。
Use this entry point for Blender tasks, alongside the task's geometry, animation and production skills. It grants no new paid, public, production, credential or customer-data authority. Do not change model routing or restart a gateway to use it.
Use a deterministic background script for a known-ahead build or long final render. For repeated inspect/edit cycles, prefer an installed, owned warm-session adapter with private pipes and isolated preferences. Load blender-warm-session when installed. Do not install an unreviewed DSH plugin, adopt an existing Blender GUI or expose a control port. If warm execution is unavailable, use the verified background-script route and disclose the fallback.
Keep one bounded resource owner, a whole-job deadline, an exclusive disposable workspace and explicit source/output identities. Preserve source files. Never detach a worker across turns. Blender exit success alone is not artifact acceptance.
Measure applicable geometry, materials, framing and lighting against the task specification. Unknown checks must be measured; failures return to the owning planner for a reviewed repair. Do not label an unperformed or inapplicable check as passed. Apply role-aware topology rules, not a blanket closed-solid assumption. Save native checkpoints before risky edits.
Use jevkit.blender.select(state) or pipe the closed JSON schema in docs/blender.md into jev blender-next. No scene files, screenshots, code, names, paths, free text or private measurements may enter this helper. The only permitted symptoms are its fixed vocabulary. private=true always bypasses inference.
Known steps run locally. Jev may prioritize a prevalidated read-only inspection only when the visual cause is unclear and a measured candidate probe costs at least one second. Cap the advice request at one second and the owned loop at two model requests and six probes. Skip Jev for cheap work. Its latency and confidence do not prove the selected cause. Preserve abstentions and failures; never lower confidence to obtain an action.
Jev cannot write Blender code, authorize a scene edit, invent arguments or declare completion. Dispatch only exact caller-owned read-only probe IDs. Verify returned results locally, then let the owning planner author any repair. Stop on stale/no-progress observations, exhausted budgets, unknown IDs or script errors. An unavailable chooser returns control to the local planner, not success.
Re-run failed checks on the repaired artifact. Inspect decoded representative views and applicable motion frames. Reopen native files and independently reimport exports where relevant. Only an explicit local artifact verifier can mark the workflow complete. Close the owned worker and verify cleanup. Report timings, failures and untested boundaries honestly. Synthetic decision tests do not prove artistic quality, fewer agent turns, universal speedups or superiority over MCP.
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Use when driving a web page in a browser — clicking, typing, navigating, logged-in or JS-rendered pages. Jev picks each step from the elements observed, under a host allowlist and a step budget.
中文简介正在准备中。当前显示原文说明。
Use when a transcript has to be cut to a fixed size and you must choose which turns go. Jev marks each turn keep, summarize or drop. Measured: it does not make a handoff better.
中文简介正在准备中。当前显示原文说明。
Use when driving a desktop GUI through a computer-use driver — windows, menus, native apps, OS dialogs. You build a table of safe actions; Jev picks the next one in about 0.4 seconds.
中文简介正在准备中。当前显示原文说明。
Use when a task is already judged hard — pick which paid frontier seat takes it, then keep Jev watching the delegated run so it interrupts you only when the run needs a decision.
中文简介正在准备中。当前显示原文说明。
Use on a mailbox export to sort mail into needs reply, updates, promotional, sales and spam — which messages are addressed to the person at all. For a support queue use jev triage.
中文简介正在准备中。当前显示原文说明。
Use on passages a search just returned (memory, vault, session history, wiki, web) before reading them in. Jev ranks them, drops the irrelevant, and flags prompt injection hidden in the text.
中文简介正在准备中。当前显示原文说明。