dev-environment
The canonical full-stack entrypoint is:
$ npx -y skills add AgentsMesh/AgentsMesh --agent claude-codeHow it fires
How this agent gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
- You can call itInvoke it directly when you want it.
Context preview
The summary Claude sees to decide when to auto-load this agent.
The canonical full-stack entrypoint is:
Agent definition
dev-environment.mdWorktree Development Environment
The canonical full-stack entrypoint is:
bazel run //deploy/dev:up
Use `//deploy/dev:backend_only` when the task does not need Web or Web Admin. Other lifecycle targets are `//deploy/dev:clean`, `//deploy/dev:reset_runners`, and `//deploy/dev:rebuild_runner`.
The environment is hybrid:
- Docker runs stateful infrastructure and containerized Runner dependencies.
- Backend, Relay, and the host Runner are launched through Bazel and ibazel.
- Web and Web Admin are Bazel Next.js development servers.
- Traefik routes the worktree's public HTTP, relay, and gRPC entrypoints.
Every worktree gets a generated `deploy/dev/.env` with isolated ports. After the stack starts:
set -a source deploy/dev/.env set +a
Read values such as `HTTP_PORT`, `WEB_PORT`, `WEB_ADMIN_PORT`, `POSTGRES_PORT`, `MINIO_API_PORT`, `MINIO_CONSOLE_PORT`, and `TRAEFIK_DASHBOARD_PORT` from that file. Do not infer them from the main checkout.
Useful logs:
deploy/dev/runtime/backend/backend.log deploy/dev/runtime/relay/relay.log deploy/dev/runtime/runner/runner.log deploy/dev/web.log
Only clean a stack that this workflow started and only after checking that no parallel task depends on it.
The AI Agent Workforce Platform. Run a hundred AI coding agents across your own machines — schedule, isolate, and steer them all from one console.
Repo: AgentsMesh/AgentsMesh
Other agents on agentsmesh.
- cochange-churn
静态切面看不到的架构问题,往往在**版本历史**里暴露。这是来自代码审查 git-history lens 的高信号视角。仅当处于 git 仓库时启用。
Open agent - concept-modeling
架构的本质是**概念的识别与组织**。这是优先级最高的 lens。
Open agent - evidence-and-detection
LLM 架构审查的头号失败模式:**凭空断言结构**——声称"A 依赖 B""这里有循环""X 被多处写",却没真去查。本文件给出强制接地原则 + 各语言的实际探测命令。
Open agent - report-format
范围小用 SKILL.md 里的精简版即可。下面是全量审计的结构化报告。
Open agent - responsibility-causality
这是本 skill 最独特、最能抓**真实结构性 bug** 的 lens:职责归属、机制/策略分离、因果与不变量(并发/状态契约)、属性三分。
Open agent - solid-grasp-yagni
经典原则做交叉验证。**这一 lens 最容易产出假阳性(教科书式建议)——严格用置信度门控,不报"理论上更优雅但无实际危害"的项。**
Open agent

