flywheel-drift-check

Detects drift between the implementation plan and current codebase state.

3|1|Updated Apr 15, 2026
One-click install
npx skills add https://github.com/burningportra/agent-flywheel-plugin --skill flywheel-drift-check-burningportra
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: flywheel-drift-check
Source: https://github.com/burningportra/agent-flywheel-plugin/tree/main/skills/flywheel-drift-check
Command: npx skills add https://github.com/burningportra/agent-flywheel-plugin --skill flywheel-drift-check-burningportra

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Long-running coding projects drift: the plan written at kickoff no longer matches what the code actually does, beads go stale, and new requirements appear that the plan never covered. This Skill compares the current codebase against the saved plan and bead graph so you can catch divergence before it derails implementation. ## Core Features & Use Cases - Plan-vs-code comparison: Reads the plan document from .pi-flywheel/checkpoint.json (or the latest file in docs/plans/) and uses an Explore agent to check which planned changes are implemented, stale, or missing. - Bead graph analysis: Runs br list --json and bv --json to surface blocked beads, dependency cycles, orphans, and bottlenecks. - Drift report with routing: Displays on-track, stale, blocked, and new-opportunity counts, then asks via AskUserQuestion whether to polish the bead graph, run a full reality-check, or ignore the drift. - Use Case: After a week of implementation, run a drift check to discover that 4 beads are stale and 2 new requirements emerged, then polish the bead graph in place instead of replanning from scratch. ## Quick Start Ask the agent to run a drift check comparing the current codebase against the saved implementation plan and bead statuses.

Frequently Asked Questions about flywheel-drift-check

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check if my codebase has drifted from the implementation plan?

Run the drift check, which reads the plan path from .pi-flywheel/checkpoint.json (or the newest file in docs/plans/), pulls bead statuses with br list --json, and uses an Explore agent to compare planned changes against actual code. It reports on-track, stale, blocked, and new-opportunity beads.

What is the difference between drift-check and reality-check?

Drift-check is the lightweight pass that detects bead-graph staleness and tactical divergence from the plan. Reality-check is the deeper strategic review for vision-versus-implementation gaps, invoked when drift suggests the plan itself no longer matches the product direction.

What happens when significant drift is detected?

When three or more beads are stale or newly identified, the skill asks via AskUserQuestion whether to polish the bead graph in place with flywheel_approve_beads, run a full reality-check, or ignore the drift and continue. Minor drift skips the escalation.

Does drift-check work without a checkpoint file?

Yes. If .pi-flywheel/checkpoint.json has no planDocument field, the skill falls back to scanning docs/plans/ for the most recent plan file, so it still works in projects that have plans but no active checkpoint.

What tools does drift-check require to analyze the bead graph?

It requires the br CLI for bead status listing (br list --json) and the bv CLI for graph analysis (bv --json), which surfaces dependency cycles, orphan beads, and bottlenecks in the work graph.