How does the skill handle task batching?

By default, the skill executes plans in batches of three tasks. After each batch is completed, it triggers a mandatory code review and reports progress to the user before continuing.

What happens if an agent cannot follow the plan?

If an agent encounters a problem that prevents it from following the plan exactly, it reports a 'BLOCKED' status. The skill then stops and asks the user to choose between trusting the agent, revising the plan, or enforcing the original approach.

How are specialized agents selected?

The skill uses semantic understanding to analyze task requirements. It selects agents like cipherpowers:rust-agent for Rust code or cipherpowers:ultrathink-debugger for complex issues, ensuring the right tool is used for the job.

Is code review mandatory during execution?

Yes, code review is an integrated and mandatory part of the workflow. The skill automatically dispatches a code-review-agent after every batch to ensure quality and plan adherence.

Executing Plans

Name: Executing Plans
Author: cipherstash

bycipherstash

•

生産性とワークフロー

Manages the systematic implementation of complex development plans through batch execution, semantic agent selection, and mandatory review checkpoints.

概要

The Executing Plans skill enables Claude to handle multi-step implementation plans with high precision and reliability by breaking down complex roadmaps into manageable batches. It ensures architectural integrity by selecting specialized agents—such as dedicated Rust agents or deep-thinking debuggers—using semantic task analysis rather than simple keyword matching. With mandatory code review checkpoints between batches and a strict 'no-deviation' policy for blockers, this skill provides a controlled and transparent environment for executing sophisticated software development tasks without losing human oversight.

主な機能

Semantic agent selection for specialized tasks like Rust coding or complex debugging
Automated development branch finalization including testing and cleanup
Strict adherence to implementation plans with mandatory 'BLOCKED' status reporting
Batch-based execution with configurable checkpoints for architect review
Integrated code review workflows using the cipherpowers:requesting-code-review sub-skill
4 GitHub stars

ユースケース

Executing multi-step refactoring plans that require precise architectural oversight
Managing long-running development tasks that require frequent human feedback and verification
Implementing complex features in polyglot codebases using specialized domain agents

概要

主な機能

Semantic agent selection for specialized tasks like Rust coding or complex debugging
Automated development branch finalization including testing and cleanup
Strict adherence to implementation plans with mandatory 'BLOCKED' status reporting
Batch-based execution with configurable checkpoints for architect review
Integrated code review workflows using the cipherpowers:requesting-code-review sub-skill
4 GitHub stars

ユースケース

Executing multi-step refactoring plans that require precise architectural oversight
Managing long-running development tasks that require frequent human feedback and verification
Implementing complex features in polyglot codebases using specialized domain agents