Execution Pipeline

The Execution Pipeline is the portion of Maudel’s pipeline where user stories are actively executed. While the full Maudel pipeline includes 12 stages (starting with Blueprint and GenStory for requirements and story generation), the Execution Pipeline encompasses the 10 stages dedicated to implementing, validating, and delivering each story.

Users can select any combination of these stages to create custom pipeline templates tailored to their project needs — with up to 10 execution stages, each with role-specific expert agents, quality gates, and artifact validation.

The Ten Execution Stages

1. Priming

Load context and prime the agent system. Pipeline instructions, stage directives, reference skills, and upstream context are assembled into a composable system prompt using the 9-layer prompt architecture.

2. GenPlan

Create the development plan. The system decomposes the work into tasks, maps dependencies, identifies test requirements, and produces a structured implementation plan that the orchestrator will execute.

3. Sandbox

Create an isolated git worktree for implementation. This ensures all changes are sandboxed, reversible, and traceable. The Pipeline Minder monitors sandbox isolation throughout execution.

4. Implement

Implementation of the plan. Code is generated, migrations are written, and configuration is applied — task by task according to the structured plan. All artifacts are registered in the traceability graph as they are produced.

5. Validate

Evaluate the implementation against acceptance criteria. The system performs architecture conformance checks, security validation, and generates QA evidence. Artifact completeness is verified against the traceability graph.

6. Unit Test

Generate and execute unit tests mapped to acceptance criteria. The system verifies code coverage thresholds are met and registers test results as traceability artifacts linked to their upstream stories.

7. Code Review

Automated code review against project standards. The system checks scope compliance, architecture alignment, and security patterns. Findings are surfaced with severity ratings and remediation recommendations.

8. GenDocs

Generate documentation from implementation artifacts. API documentation, inline comments, and changelog entries are produced and verified to trace back to upstream intent in the traceability graph.

9. SelfLearn

Extract learnings from the execution run. The system updates agent skills and reference knowledge, feeding patterns into cross-execution improvement loops for continuous refinement.

10. Merge

Validate that worktree changes merge cleanly to the target branch. A final quality gate evaluates all artifacts across the entire execution run. Merge executes with a full audit trail and traceability linkage.

Quality Gates

Each stage transition is gated by the deterministic orchestrator. A gate evaluates signals including:

• Schema validation — does the artifact conform to its expected structure?
• Completeness checks — are all required fields and sections present?
• AI rubric scoring — does the output meet quality thresholds?
• Dependency resolution — are upstream requirements satisfied?
• Test results — do generated tests pass?

Gate decisions are deterministic: accept, revise, retry, escalate, or block.

Resumability

The pipeline uses a checkpoint-based state machine. If execution fails at any stage, work can resume from the failed step with full state consistency. States include: PENDING, RUNNING, COMPLETED, FAILED, WAITING_HITL, PAUSED, and ABORTED.