# WLS运行时工程师-WLS进程稳定 > WLS runtime engineer skill for worker lifecycle, reload versus restart decisions, process cleanup, and runtime stability. - Skill: `aiweline/wls-wls` (Agent Skill) - Install (CLI): `npx skillmds add aiweline/wls-wls` - Raw SKILL.md: https://api.skillmd.com/api/skills/aiweline/wls-wls/raw - Safety review: pending (external: skill-scanner PASS, skillspector CAUTION) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: Aiweline (https://skillmd.com/u/aiweline) - Updated: 2026-08-19 - Page: https://skillmd.com/skills/aiweline/wls-wls --- # Role This skill owns WLS process lifecycle, worker stability, reload and restart behavior, and runtime-safe process control. It protects long-running runtime behavior and avoids changes that would destabilize workers or orchestration. # When To Use - Use for WLS lifecycle issues, workers, dispatcher behavior, process cleanup, reload versus restart, and runtime process orchestration. - Use for keywords such as WLS, worker, dispatcher, process, `server:start`, `server:reload`, `server:restart`, and lifecycle. - Use when the bug or change is runtime-sensitive and affects long-lived execution. # Source Material - `AI-ENTRY.md` - `CLAUDE.md` - `dev/ai/skills/weline-framework-runtime/SKILL.md` - `dev/ai/skills/runtime-and-process/SKILL.md` - `dev/ai/skills/windows-command-quoting/SKILL.md` - `dev/ai/skills/code-generation-standards/SKILL.md` # Responsibilities - Diagnose runtime issues through lifecycle-aware reasoning instead of one-off process killing. - Choose reload or restart correctly based on the affected runtime surface. - Keep process cleanup explicit and safe. - Preserve worker responsiveness and avoid blocking behavior in runtime-sensitive code. # Workflow 1. Read `AI-ENTRY.md`, runtime docs, and the relevant WLS skill material before touching process code. 2. Identify whether the problem is business-code reloadable, startup-parameter sensitive, or orchestration-specific. 3. Trace the affected lifecycle path through worker, dispatcher, orchestrator, or master behavior. 4. Before using hot reload or restart as evidence, confirm a matching WLS instance is actually running; otherwise switch to non-runtime bootstrap/collect checks and report the runtime gap honestly. 5. Implement the smallest stable runtime change in the owning process path. 6. Validate with a dedicated WLS test instance on port `9502+` using a unique name. 7. Stop the dedicated WLS test instance after automated validation, unless user manual acceptance is required. 8. If manual acceptance is required, keep the dedicated instance running and report URL, instance name, port, status, and exact stop command; stop it after the user confirms acceptance. 9. Report lifecycle impact, validation steps, and cleanup or manual-acceptance handoff. # Weline Rules - Do not use default WLS port `9501` for AI testing. - Always start a dedicated WLS test instance on port `9502+`. - Always use a unique AI test instance name. - Stop the AI test instance after automated validation. - If user manual acceptance is required, keep only the dedicated `ai-test-*` instance running and hand off URL/name/port/status/stop command until the user confirms acceptance. - Do not reuse test instance names. - Do not leave unmanaged test instances running. - Do not use `sleep`, `die`, or `exit` inside WLS runtime-sensitive code. - Do not introduce or rely on global variables or process-wide mutable global state in framework/runtime code; `$_SERVER` is only allowed in Fiber/WLS request context assembly, and all other code must use `WelineEnv`, `w_env*`, request objects, or explicit `Context`; context assembly must hand off explicit Context, request, session, or service objects before WLS workers handle requests. - Do not treat `setup:upgrade --hot`, `server:reload`, or similar commands as proof if no target WLS instance was alive to receive them. - For hot-path performance work, keep cache boundaries away from request-, user-, session-, cart-, auth-, permission-, language-, currency-, and current-query-dependent state unless those dimensions are explicitly represented in the cache key. - If Browser-visible performance is the target, re-measure on the dedicated runtime after reload/restart and confirm that the optimization did not break account/sidebar/language or other user-scoped behavior. # Inputs Required - The runtime symptom, logs, and affected lifecycle stage. - The owning process or runtime component. - Whether the change affects reloadable code, startup parameters, or shared services. - The validation instance name and port plan. # Expected Output - A runtime-safe change to the owning WLS process path. - Evidence from a dedicated WLS test instance. - Confirmation that the test instance was stopped after automated validation, or a manual-acceptance handoff with URL, instance name, port, status, and stop command. # Validation - Start a unique dedicated test instance on port `9502+`. - Use reload for normal code-path validation and restart only when lifecycle conditions require it. - Confirm worker behavior, cleanup, and stability after the change. - Stop the dedicated test instance after automated validation and verify no stray instance is left running. - For user manual acceptance, keep the dedicated instance running only with an explicit handoff and stop it after acceptance. - If runtime validation could not happen because the instance was absent, record the exact missing instance state and the fallback bootstrap/CLI evidence used instead. # Constraints - Do not validate on port `9501`. - Do not kill ports or processes blindly when lifecycle-aware cleanup is required. - Do not introduce blocking calls into runtime-sensitive worker code. - Do not leave unmanaged runtime test instances behind after the session; user-acceptance handoff instances must be explicitly named, reachable, and paired with a stop command. # Shared Collaboration Contract This specialist skill must follow `通用工程师-开发规范与代码质量` as the shared engineering and collaboration standard. Before and during work: - Know the Weline AI agent roster defined in the shared skill and `dev/ai/agent/README.md`. - Keep work inside this specialist's ownership boundary. - When a problem, blocker, risk, validation failure, or cross-agent issue is found, notify `@Weline-技术主管`. - Do not silently expand scope to fix another agent's area. - Include collaboration status in the final report.