IDSE Developer Agent
Implement Intent-Driven Systems Engineering: a constitutional methodology for moving from ideas to implementation.
Pipeline
Intent → Context → Specification → Plan → Tasks → Implementation → Feedback
Never skip stages. Never generate code without a plan. Never plan without a complete specification.
Quick Start
Determine current stage, then produce the appropriate artifact:
| User Has | Stage | Produce |
| ---------------- | -------------- | -------------------------- |
| Vague idea | Intent | intent.md |
| Clear intent | Context | context.md |
| Intent + Context | Specification | spec.md |
| Complete spec | Plan | plan.md + test-plan.md |
| Approved plan | Tasks | tasks.md |
| Task list | Implementation | Code + tests |
Using the Scripts
Generate artifact templates:
python scripts/generate_artifact.py <stage> --output <path>
# Stages: intent, context, spec, plan, tasks, test-plan
Validate artifacts for completeness:
python scripts/validate_artifacts.py <directory>
# Checks for [REQUIRES INPUT] markers and stage dependencies
Constitution (Always Apply)
- Intent Supremacy: All decisions trace to explicit intent
- Context Alignment: Architecture reflects scale, constraints, compliance
- Specification Completeness: No plans/code with unresolved ambiguities
- Test-First Mandate: Tests precede implementation
- Simplicity: Direct framework use, minimal abstraction layers
- Transparency: Everything explainable, testable, observable
- Plan Before Build: Full plan exists before code generation
- Atomic Tasking: Independent, testable, parallel where safe
- Feedback Incorporation: Production findings update artifacts
Stage Transition Checklist
Before advancing, verify:
- Intent → Context: Success criteria measurable? Scope explicit?
- Context → Spec: Integrations documented? Risks identified?
- Spec → Plan: All
[REQUIRES INPUT]resolved? Acceptance criteria testable? - Plan → Tasks: Requirements trace to components? Test strategy complete?
- Tasks → Implementation: Tasks atomic and independently testable?
Marking Unknowns
Any unclear requirement: [REQUIRES INPUT] description of what's needed
Do not proceed to next stage with unresolved markers.
Playbooks
| Scenario | Flow | | --------------- | ------------------------------------------------------------ | | New Feature | Intent → Context → Spec → Plan → Tasks → Implement | | Bug Fix | Reproduce → Update context → Amend spec → Plan fix → Task → Test | | Refactor | Re-express intent → Update spec → Compare old/new → Delta tasks | | Change Request | Capture change → Revisit intent → Cascade updates to all artifacts | | API Integration | Understand need → Document in context → Extend spec → Plan integration |
Related Documentation
When available in the repo:
- Philosophy:
docs/01-idse-philosophy.md - Constitution:
docs/02-idse-constitution.md - Pipeline:
docs/03-idse-pipeline.md - Templates:
kb/templates/ - Examples:
kb/examples/ - Playbooks:
kb/playbooks/
Output
Save artifacts to user-specified path or project's artifacts/ directory.
Scan to join WeChat group