菜单
Persistent code-mode kernels (JS/Python) for incremental computation — omp/senpi eval analog on GrokBuild via lfg MCP. Use when multi-step code execution, data exploration, or stateful REPL beats one-shot bash.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Execute a Prometheus work plan in GrokBuild with `/goal` state, Boulder state, evidence ledger updates, worktree discipline, parallel subagents, and explicit continuation. Use after planning when the user says start work, execute plan, continue plan, resume plan, or asks to run a .omo/plans plan.
Goal-like loop that uses ultrawork mode to decompose work into systematic, evidence-bound steps.
MUST USE for planning before coding: 5+ steps, ambiguous scope, multiple modules, architecture decisions, a vague 'just make it good / figure out what to build' brief, or any request to plan, interview, or break work down. Explore-first planning consultant (Prometheus) that grounds in the codebase, asks only the forks exploration cannot resolve - or researches them to best practice when the intent is fuzzy - waits for explicit approval, then writes ONE decision-complete work plan a worker executes with zero further interview. Triggers: ulw-plan, plan this, make a plan, plan before coding, interview me, break this down, start planning, plan mode, just make it good, figure out what to build.
Execute a Prometheus work plan in GrokBuild with `/goal` state, Boulder state, evidence ledger updates, worktree discipline, parallel subagents, and explicit continuation. Use after planning when the user says start work, execute plan, continue plan, resume plan, or asks to run a .omo/plans plan.
Goal-like loop that uses ultrawork mode to decompose work into systematic, evidence-bound steps.
Post-implementation review orchestrator. Launches 5 parallel background sub-agents: Oracle (goal/constraint verification), Oracle (code quality), Oracle (security), unspecified-high (hands-on QA execution), unspecified-high (context mining from GitHub/git/Slack/Notion). All must pass for review to pass. MUST USE after completing any significant implementation work. Triggers: 'review work', 'review my work', 'review changes', 'QA my work', 'verify implementation', 'check my work', 'validate changes', 'post-implementation review'.
| name | eval |
| description | Persistent code-mode kernels (JS/Python) for incremental computation — omp/senpi eval analog on GrokBuild via lfg MCP. Use when multi-step code execution, data exploration, or stateful REPL beats one-shot bash. |
Grok-adapted code mode inspired by oh-my-pi eval and senpi senpi-codemode.
Use the eval MCP server tools (installed by lfg setup --run):
| Tool | Purpose |
|---|---|
eval | Run one cell in a persistent kernel (js or py) |
eval_reset | Wipe kernel state for a language or all |
Prefer eval for:
Prefer bash only for short single pipelines (ls, git status, one-shot CLI).
One eval call = one logical step.
Do not re-import / re-declare names already defined unless you reset, the kernel crashed, or you got NameError / ReferenceError.
{
"language": "js" | "py",
"code": "<cell body, verbatim>",
"title": "optional label",
"timeout": 30,
"reset": false,
"cwd": "/optional/workdir"
}
reset: true wipes that language only before the cell.(cwd, language) for the life of the MCP process.| Token | Runtime |
|---|---|
js | Node VM worker (top-level await ok) |
py | python3 subprocess (exec in persistent globals) |
Ruby/Julia kernels exist in omp/senpi but are not in this Grok MVP.
Python — incremental
eval language=py code="import json" title=imports
eval language=py code="data = {\"n\": 3}" title=define
eval language=py code="print(data[\"n\"] * 2)" title=use
JavaScript — top-level await
eval language=js code="const n = 2" title=define
eval language=js code="n + 40" title=use
Reset after poison state
eval_reset language=py
ok: false with traceback; prior names still alive.python3 → py cells fail until available (override with LFG_EVAL_PYTHON).tool.* / agent() / completion() bridges (omp full eval has those).