Plan Execution Orchestrator
Execute a phased implementation plan using subagents, with built-in verification and gated commits.
The skill instructs orchestrating subagents for task execution with verification and commit approval, but does not limit subagent permissions or disclose data flow, nor mention sensitive data handling. Repository-level SECURITY.md shows good practices, but the skill itself lacks runtime user confirmation and isolation. Deducted for incomplete permissions and confirmation, scored 12.
The skill instructions are internally consistent with clear flow, but no runnable tests or examples provided, depends on external plan files, and abnormal input behavior undefined. Static review cannot verify key paths, but no contradictions found, scored 10 due to lack of tests and error handling.
Skill fits 'execute plan' scenarios, but trigger conditions rely only on description keywords with vague boundaries, no non-fit scenarios declared. Environment: main repo supports Chinese modes, but skill itself does not mention Chinese or mainland accessibility, relies on external services. Scored 7 for clear scenario but limited boundary/trigger evidence.
Skill docs are clearly structured with execution protocol and failure modes, but lacks install/dependency notes, parameter examples, known limitations, and version history. Repo has license and author, but skill file itself lacks maintenance responsibility. Scored 10 for readability but incomplete governance.
Skill aims to execute plans efficiently via subagents, but lacks actual use cases or output verification, value claim unverified. Static review cannot confirm task completion correctness, marginal value limited. Scored 7 for core task feasible but evidence insufficient.
Repo has CI workflows and test suites, but tests for the skill file are not explicit, and static review cannot execute. The orchestration flow claims lack independent verification. Scored 5 for auditable material but thin coverage.
- When executing, the skill delegates to subagents; ensure subagents have least privilege and do not over-authorize.
- Skill depends on plan files; verify the source and feasibility of the plan content.
- Repository relies on external services (e.g., model providers) that may be inaccessible from mainland China; check network reachability or prepare alternatives.
- Skill lacks data-flow and sensitive-data handling disclosure; review subagent behavior when using.
- Static assessment cannot verify real execution; recommend a pilot run.
What does this skill do, and when should you use it?
This skill is designed to carry out a phased implementation plan, especially one created by make-plan. It instructs an orchestrator to deploy subagents for all execution work, rather than doing the work itself. After each phase, it mandates separate verification, anti-pattern, and code-quality review subagents, and only allows commits after verification passes. The skill explicitly targets common failure modes like inventing APIs, skipping verification, and committing prematurely.
The skill defines an Execution Protocol: each phase deploys an 'Implementation' subagent to execute as specified and copy patterns from documentation, citing sources for unfamiliar APIs. After each phase, it deploys a 'Verification' subagent to prove the phase works, an 'Anti-pattern' subagent to grep for known bad patterns, a 'Code Quality' subagent to review changes, and a 'Commit' subagent only after verification passes. Between phases, a 'Branch/Sync' subagent pushes to a working branch and prepares the next handoff. Subagents must report evidence such as commands run, outputs, and files changed.
- A developer has a detailed implementation plan from make-plan and wants it executed reliably with verification and review at every step.
- A tech lead wants to delegate a large refactoring effort to subagents while retaining control over verification and commit discipline.
- A team working with unfamiliar APIs wants to enforce that code follows documented usage rather than speculative API calls.
- A programmer wants to automate the common dev workflow of implement-verify-commit in a phased manner with anti-pattern detection.
What are this skill's strengths and limitations?
- Automates phased execution, reducing manual coordination.
- Built-in verification, anti-pattern, and code-quality checks help prevent errors.
- Blocks commits until verification passes, promoting safer development.
- Emphasizes using documented API patterns, reducing API misuse.
- Requires a platform capable of spawning and coordinating subagents; standard model APIs won't work as-is.
- Specifics of the verification checklist are not defined; implementation is left to the user.
- No test suite or cross-platform testing is mentioned; reliability on other clients is unverified.
- It does not perform work itself, relying on the underlying platform's subagent support.
How do you install this skill?
The skill lives in the open-source repository thedotmack/claude-mem at path openclaw/skills/do/SKILL.md. The repository is a monorepo containing multiple skills, and the README provides installation instructions for the whole collection (e.g., npx claude-mem install or via the plugin marketplace). To install just this skill, copy the 'do' folder containing SKILL.md into your agent's skills directory. No standalone installation steps are provided for this skill alone.
How do you use this skill?
Trigger this skill when you ask to execute, run, or carry out a plan — especially one created by make-plan. The orchestrator will follow the protocol in SKILL.md to deploy subagents. Example trigger prompt: 'Use the do skill to execute the current plan.' The skill will guide the orchestrator through deploying Implementation, Verification, Anti-pattern, Code Quality, Commit, and Branch/Sync subagents after each phase.
How does this skill compare with similar options?
This skill focuses on orchestrating multi-agent implementation workflows. No alternatives are named in the source, so no direct comparison is provided.