Completion Verification
npx machina-cli add skill a5c-ai/babysitter/completion-verification --openclawCompletion Verification
Verify all phases are complete with weighted quality scoring before allowing session exit.
Agent
Completion Verifier - pwf-completion-verifier
Workflow
- Read all three planning files in parallel
- Parse checkbox completion state from task_plan.md
- Cross-reference plan, findings, and progress for consistency
- Check error resolution status
- Calculate weighted quality score (phases 40%, errors 25%, findings 20%, continuity 15%)
- Generate recommendations if threshold not met
- Write verification report to progress.md
Inputs
projectPath- Root path for planning filestaskDescription- Original task descriptionqualityThreshold- Minimum score to pass (default: 80)strictMode- Require 100% checkbox completion (default: false)
Outputs
- Quality score with component breakdown
- Phase completion report
- Unresolved error list
- Prioritized improvement recommendations
Process Files
planning-orchestrator.js- Completion assessment and final verificationplanning-verification.js- Full verification pipeline
Source
git clone https://github.com/a5c-ai/babysitter/blob/main/plugins/babysitter/skills/babysit/process/methodologies/planning-with-files/skills/completion-verification/SKILL.mdView on GitHub Overview
Completion Verification ensures all planning phases are complete and the session cannot exit until quality criteria are met. It reads three planning files in parallel, validates checkbox completion in task_plan.md, and cross-references plans, findings, and progress to guarantee consistency before scoring.
How This Skill Works
The agent reads all three planning files in parallel, parses the checkbox completion state from task_plan.md, and cross-references plan, findings, and progress for consistency. It then checks error resolution status, computes a weighted quality score (phases 40%, errors 25%, findings 20%, continuity 15%), and generates recommendations if the threshold isn’t met, finally writing a verification report to progress.md.
When to Use It
- Before exiting a planning session to ensure all tasks are complete
- When coordinating multiple contributors across planning files
- When an objective pass/fail gate is needed based on a quality threshold
- To surface and list unresolved errors from planning and progress reviews
- To auto-generate prioritized improvement recommendations and document them
Quick Start
- Step 1: Set inputs: projectPath, taskDescription, qualityThreshold (default 80), strictMode (default false)
- Step 2: Run the Completion Verifier agent pwf-completion-verifier
- Step 3: Open progress.md to review the quality score, phase report, unresolved errors, and recommendations
Best Practices
- Run the verifier after all planning files have been updated to ensure current data
- Enable strictMode only when you require 100% checkbox completion before exit
- Regularly review the cross-reference results across plan, findings, and progress
- Tweak the weighted breakdown (phases 40%, errors 25%, findings 20%, continuity 15%) to match project priorities
- Always store the verification report in progress.md for auditability
Example Use Cases
- A product team uses the verifier to gate release after completing all planning phases and resolving critical errors
- An AI assistant project ensures findings align with planned milestones before progressing to implementation
- A multi-contributor planning effort surfaces inconsistencies and highlights unresolved errors
- The tool outputs a prioritized improvement list to guide subsequent planning cycles
- The agent automatically writes a progress.md report documenting the quality score and component breakdown
