gaoguobin/codex-eide-rebuild / skills/eide-rebuild
eide-rebuild
VerifiedEIDE rebuild and compile Agent Skill for Embedded IDE for VS Code workspaces. Runs a Python unify_builder runner and returns compact or full JSON build results.
Use this skill when the user wants an EIDE project rebuilt from Codex.
Trigger patterns
- Natural language build requests such as
你自己编译验证下对不对 - Natural language build requests such as
帮我编译确认一下 - Natural language build requests such as
先 rebuild 看结果 - Explicit requests such as
EIDE rebuild C:\work\demo\project.code-workspace - Long-log requests such as
EIDE subagent rebuild C:\work\demo\project.code-workspace
How to execute
- Resolve the target path.
- Prefer a user-provided
.code-workspacepath. - If the user points at a project directory, pass that directory to the runner and let the runner resolve the single workspace file.
- If no path is supplied, inspect the current working directory. Use it when it contains exactly one
.code-workspacefile. - If the current working directory does not identify one workspace, ask the user for the path.
Decision table:
| Request shape | Default execution |
|---|---|
| Single project, normal compile/rebuild request | Run the runner directly in the main agent. |
| Single project, explicit subagent/delegated/background request | Use the eide-rebuild custom agent when available. |
| Multiple independent projects, explicit subagent/parallel/delegated request | Use one rebuild worker per project when the host supports delegation. |
| Same project or same build directory | Never run rebuild workers concurrently. |
Run:
python scripts/eide_rebuild.py rebuild <workspace-or-project-path> --stdout minimal
The relative scripts/eide_rebuild.py path is relative to this installed skill directory. Host integrations can use their own installed absolute runner path.
Environment check:
python scripts/eide_rebuild.py doctor
Result handling
- Treat the runner as the source of truth.
- Prefer
--stdout minimalfor normal agent work. It prints bounded JSON status and still writes the complete result toresultPath. - Use
--stdout summaryonly when stdout needs per-target artifacts, memory, source stats, failures, or diagnostics without openingresultPath. - Use
--stdout fullor omit--stdoutonly when the full JSON is explicitly needed on stdout. - First inspect
ok,exitCode,errorCode,summary,targetNames,targets[].ok,targets[].failureCount,targets[].diagnosticCount,targets[].artifactCount, andresultPath. - On success, report only a low-noise summary by default:
ok,exitCode,errorCode,summary,targetNames, failure count, diagnostic count, artifact count, andresultPath. - Use
targets[].artifacts[].sha256as artifact identity/provenance data. Do not treat hash changes across different rebuilds as success/failure evidence unless the user explicitly asks for reproducibility or deterministic build checks. - Keep the complete JSON available at
resultPath; it includescompilerLog,steps,artifacts, andtranscript. - Do not paste full logs or long artifact lists on successful builds.
- Read
compilerLog,transcript, andsteps[].stdout/stderrfromresultPathonly when the build fails, the user asks for details, or the structuredfailures/diagnosticsfields are not enough. - Use exit code
0for success,6for build failure, and the other exit codes for environment or tool errors.
Subagent guidance
- Prefer the
eide-rebuildcustom agent when the user explicitly asks for subagent/delegated rebuild work and the host supports delegation. - Otherwise use direct runner execution for normal single-project requests.
- Use generic worker subagents only when the host supports delegation, policy allows it, and the user explicitly asks for subagent/parallel/delegated/background work.
- The worker should run the same Python runner with
--stdout minimal. - The worker should return only the minimal stdout plus a short factual conclusion, not paste the full result JSON, full logs, or long artifact lists.
- The main agent should keep
resultPathavailable for follow-up analysis. - Do not run multiple rebuild workers against the same project/build directory concurrently.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 468k | Updated | Claude CodeCodex | Unverified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |