What happens if a step fails verification?

Failure handling stops immediately, investigates the issue, proposes a fix, and requires explicit approval for any deviations.

Do I need to run code and security reviews?

Yes. Run code-reviewer and security-reviewer agents; the process halts if the security review fails.

How are stakes enforced during execution?

High stakes halt without a plan, medium stakes prompt a question, and low stakes may proceed inline.

plan-implementation

Scanned

npx machina-cli add skill a5c-ai/babysitter/plan-implementation --openclaw

Files (1)

SKILL.md

1.8 KB

Plan Implementation

Overview

Execute approved plans with discipline. Each step follows a read-modify-verify loop. Phase checkpoints require human approval. Failures trigger investigation before proceeding.

When to Use

After a plan has been approved via the planning phase
For medium and high stakes changes (low stakes can proceed inline)
When structured execution with verification is needed

Process

Load plan - Validate plan exists and is approved
Stakes enforcement - High: halt without plan. Medium: ask. Low: proceed.
Worktree isolation - Offer git worktree based on stakes level
Execute steps - For each: mark in-progress, locate files, read, modify, verify, mark complete
Phase checkpoints - Summarize and ask: continue, review, or pause
Failure handling - Stop immediately, investigate, propose fix, get approval for deviations
Code review - Run code-reviewer agent (APPROVE / APPROVE_WITH_NITS / REQUEST_CHANGES)
Security review - Run security-reviewer agent (halt if failed)
Completion summary - Steps completed, files changed, test status, plan location

Key Rules

Follow the plan strictly; deviations require explicit approval
Verify before declaring done; run verification after each step
Track progress visibly via plan document updates
Read files before modifying them
Complete code and security reviews before finishing

Tool Use

Invoke via babysitter process: methodologies/rpikit/rpikit-implement

Source

git clone https://github.com/a5c-ai/babysitter/blob/main/plugins/babysitter/skills/babysit/process/methodologies/rpikit/skills/plan-implementation/SKILL.md

View on GitHub

Overview

Plan implementation executes approved plans with a read-modify-verify loop, phase checkpoints, and mandatory reviews. It ensures failures are investigated before proceeding and requires human approval at critical steps.

How This Skill Works

Load the approved plan, enforce stakes, and isolate work with a git worktree if needed. Each step is read, modified, verified, then marked complete; phase checkpoints prompt continuation or review, and failures trigger investigation before any deviation.

When to Use It

After a plan is approved via the planning phase.
For medium or high-stakes changes where governance matters.
When structured execution with step-by-step verification is required.
When phase checkpoints require human approval before advancing.
When code and security reviews must be completed before finishing.

Quick Start

Step 1: Load plan, validate existence and approval.
Step 2: Execute each step with read-modify-verify and update status.
Step 3: Run code and security reviews, then complete the summary and reference plan location.

Best Practices

Follow the plan strictly; deviations require explicit approval.
Verify before declaring done; run verification after each step.
Track progress visibly via plan document updates.
Read files before modifying them.
Complete code and security reviews before finishing.

Example Use Cases

Rolling out a security-sensitive patch with mandatory code and security reviews.
Medium-stakes feature rollout using a worktree to isolate changes.
Fail-fast investigation is triggered when a step cannot verify its outcome.
Pausing at phase checkpoints for manager sign-off before proceeding.
Final completion summary documenting changes, test results, and plan location.

Frequently Asked Questions

Add this skill to your agents