菜单
Use when configuring Codex CLI, OpenClaw, or scheduled agent wakes for vibe12_agent_spec. Triggers on: Codex CLI, OpenClaw, agent wake, vibe12_agent_spec, BUILD_CHECKPOINTS, memory bootstrap, cron agent.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
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。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/bbartling/py-bacnet-stacks-playground --skill vibe12-agent-runner命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Calibrate an existing-building EnergyPlus model against measured energy and operational evidence using auditable Guideline-14-style scorecards.
Build or modify a defensible EnergyPlus seed model from building evidence, including geometry, schedules, HVAC mapping, and controlled IDF changes.
Route shared EnergyPlus work to the right intake, authoring, calibration, tariff, results, or DSM procedure.
基于 SOC 职业分类
正在显示 SKILL.md
| name | vibe12-agent-runner |
| description | Use when configuring Codex CLI, OpenClaw, or scheduled agent wakes for vibe12_agent_spec. Triggers on: Codex CLI, OpenClaw, agent wake, vibe12_agent_spec, BUILD_CHECKPOINTS, memory bootstrap, cron agent. |
Automation is optional. The repo works with interactive Cursor/Codex without cron.
You are working on vibe_code_apps_12 (Vibe12 BACnet → AWS → BRICK/FDD).
Read in order:
1. vibe12_agent_spec/AGENTS.md
2. vibe12_agent_spec/scratch/memory-bootstrap-latest.md (or run bin/vibe12_workspace_cli.sh memory bootstrap)
3. vibe12_agent_spec/BUILD_CHECKPOINTS.md — execute ONE slice only
4. vibe12_agent_spec/GUARDRAILS.md
Use skills under vibe12_agent_spec/skills/ for cloud API, Ansible, BRICK, FDD, pcap.
Human owns SSH and points.csv enablement. Verify with ./scripts/validate_cloud_pipeline.sh before claiming done.
cd ~/py-bacnet-stacks-playground/vibe_code_apps_12
vibe12_agent_spec/bin/vibe12_workspace_init.sh # local MEMORY + PHASE_NOTEPAD (not in Git)
codex login
| Role | Model | Output |
|---|---|---|
| Mini | gpt-5.4-mini | Code/docs — one Next for mini slice per invocation |
| Critique | gpt-5.5 | Rewrites Last critique, Next for mini (ordered) — minis follow that queue |
Context each wake: cron_codex/state/context_since_last_wake.md (from operator_notes.md + PHASE_NOTEPAD).
vibe12_agent_spec/bin/vibe12_codex_tui.py # interactive
vibe12_agent_spec/cron_codex/bin/vibe12_wake.sh # N minis + critique
MINI_INVOCATIONS_PER_WAKE=1 vibe12_agent_spec/cron_codex/bin/vibe12_wake.sh
cp vibe12_agent_spec/cron_codex/env.example vibe12_agent_spec/cron_codex/.env
# Cron (human only — requires --yes; see GUARDRAILS)
vibe12_agent_spec/cron_codex/bin/vibe12_install_cron.sh --yes
Full guide: docs/agent-getting-started.md
vibe12_agent_spec/bin/vibe12_workspace_cli.sh memory list
vibe12_agent_spec/bin/vibe12_workspace_cli.sh memory search telemetry
vibe12_agent_spec/bin/vibe12_workspace_cli.sh memory bootstrap > vibe12_agent_spec/scratch/memory-bootstrap-latest.md
Same layout as bas_build_spec: curated MEMORY.md, daily notes, domain files under memory/, promotion after critique. No secrets in Markdown.
vibe_code_apps_11/bas_build_spec is the prior BAS app orchestration. Vibe12 uses vibe12_agent_spec/ — do not mix checkpoint queues.