Neura Market
Paul logo

Paul

Free

Plan-Apply-Unify Loop — Structured AI-assisted development for Claude Code. Quality over speed-for-speed's-sake.

FreeFree tier
#JavaScript
Type
Saas
Visit Website GitHub (864 stars)

About Paul

PAUL (Plan-Apply-Unify Loop) is a structured development methodology for AI-assisted coding with Claude Code, designed to extend GSD (Get Shit Done) by adding an enforced reconciliation phase. It focuses on quality, consistency, and full traceability by ensuring every plan is followed by an APPLY phase and then a mandatory UNIFY phase that produces a SUMMARY.md capturing actual outcomes, decisions, issues deferred, and acceptance criteria results. PAUL optimizes token-to-value efficiency by executing in the current context by default, reserving parallel subagents only for bounded research tasks. It reduces decision fatigue by suggesting a single best next action based on project state and enables zero-context session resumption through structured HANDOFF-{date}.md files. Acceptance criteria are treated as first-class citizens.

Key Features

Explicit Plan-Apply-Unify Loop with enforced UNIFY phase
Context-aware execution optimizing token-to-value efficiency
Single best next action recommendation based on project state
Structured session continuity with HANDOFF-{date}.md files for zero-context resumption
Acceptance criteria treated as first-class citizens
SUMMARY.md capturing decisions, issues deferred, and acceptance criteria pass/fail

Pros & Cons

Pros
  • Ensures complete traceability and audit trail through UNIFY summaries
  • Optimizes token usage for cost efficiency by avoiding wasteful subagent spawning
  • Reduces decision fatigue by recommending a single next action
  • Enables reliable zero-context resumption between sessions
  • Focuses on delivering high-quality results consistently
Cons
  • Requires discipline to consistently follow the UNIFY reconciliation phase
  • May feel slower for simple or one-off tasks due to enforced structure
  • Designed primarily for Claude Code, limiting applicability to other AI tools

Best For

AI-assisted software development with Claude CodeLong-running development sessions requiring full traceabilityCollaborative coding projects needing reliable context resumptionReducing decision fatigue in AI-driven workflowsProjects where quality and audit trails are prioritized over raw speed

FAQ

What is the UNIFY phase in PAUL?
The UNIFY phase is a mandatory reconciliation step that produces a SUMMARY.md document capturing what actually happened versus what was planned, decisions made during execution, issues deferred for later, and acceptance criteria results (pass/fail).
How does PAUL differ from GSD?
PAUL builds on GSD by adding an enforced UNIFY phase, executing in the current context by default to optimize token efficiency, suggesting a single next action instead of multiple options, and providing structured handoff files for reliable session continuity.
Is PAUL free to use?
Yes, PAUL is an open-source methodology hosted on GitHub under the repository of ChristopherKahler/paul, and appears to be freely available.