openclaw-workspace-2026/CONTEXT.md

CONTEXT.md

Software Development Team

Status: Active - Team creation in progress
Last Updated: 2026-07-03

Purpose

A dedicated multi-agent software development team for Beawit's internal applications and automation tooling.

Team Structure

Agent ID	Role	Responsibilities	Model
dev-product	Product Manager	Requirements, user stories, feature prioritization, acceptance criteria	kimi-k2.5:cloud
dev-architect	Architect	System design, API design, data models, tech decisions	kimi-k2.5:cloud (reasoning)
dev-backend	Backend Developer	APIs, databases, integrations, business logic	qwen3-coder:480b-cloud
dev-frontend	Frontend Developer	UI/UX, components, client state, styling	qwen3-coder:480b-cloud
dev-qa	QA Engineer	Test plans, automated tests, bug reproduction	glm-4.7:cloud
dev-devops	DevOps Engineer	Deployments, CI/CD, monitoring, infrastructure	kimi-k2.5:cloud
dev-lead (me)	Tech Lead	Coordination, code review, integration, delegation	kimi-k2.5:cloud

Workspace Organization

Directory Structure

workspace/
├── Core Config (keep in root)
│   ├── IDENTITY.md, SOUL.md, AGENTS.md
│   ├── MEMORY.md, CONTEXT.md, PROJECTS.md
│   ├── USER.md, TOOLS.md, HEARTBEAT.md
│   └── ARCHITECTURE.md
│
├── docs/           # Documentation files
│   ├── summaries/  # *SUMMARY.md files
│   ├── plans/      # *PLAN.md files
│   ├── inventories/ # *INVENTORY.md files
│   └── *.md        # Other documentation
│
├── scripts/        # Executable scripts
│   ├── checks/     # check-*.sh scripts
│   ├── fixes/      # fix-*.sh, fix-*.js
│   └── utils/      # get-*, update-*, debug-*
│
├── tests/          # Test and debug files
├── runtime/        # Package and state files
├── architecture/   # Architecture components
├── memory/         # Memory system files
└── Projects/       # Project directories

Organization Rules

1. Core config stays in root - Never move IDENTITY.md, SOUL.md, etc.
2. New documentation goes to docs/ - Sort by type (summaries, plans, inventories)
3. New scripts go to scripts/ - Categorize by purpose (checks, fixes, utils)
4. Test files go to tests/ - Any test-* or debug-* files
5. Runtime files go to runtime/ - package.json, state files, binaries
6. Run organizer monthly - bash ~/.openclaw/skills/workspace-organizer/scripts/organize-workspace.sh --dry-run

Maintenance

• Use WORKSPACE_ORGANIZATION_LOG.md to track changes
• Run dry-run before executing organization
• Archive old fix scripts after 30 days
• Clean temp files weekly

---

Technology Stack

• Backend: Python (FastAPI) — chosen for type hints, async support, and your existing Python codebase
• Frontend: HTMX + Jinja2 templates — lightweight, server-rendered, minimal JavaScript complexity
• Database: PostgreSQL for production, SQLite for local dev
• Infrastructure: Docker containers, VPS deployment via SSH
• Version Control: Git with feature branch workflow

Workflow

1. Intent Classification → Orchestrator classifies request using workflow-router
2. Context Building → Preprocessing pipeline loads rules, preferences, memory
3. Feature Request → dev-product creates user stories + acceptance criteria
4. Design → dev-architect designs system changes, API contracts
5. Implementation → dev-backend + dev-frontend work in parallel
6. Review → dev-lead reviews code, requests changes if needed
7. Validation → Automated checks via validation layer + format locker
8. Testing → dev-qa validates against acceptance criteria
9. Deploy → dev-devops deploys to staging, then production
10. Verify → dev-lead confirms deployment success

Architecture Components

Component	File	Purpose
Orchestrator	architecture/orchestrator.js	Main integration point
Pipeline	architecture/pipeline.js	Context preprocessing
Workflow Router	architecture/workflow-router.js	Intent classification
Validator	architecture/validator.js	Response quality checks
Format Locker	architecture/format-locker.js	Output template enforcement

Usage Examples

# Classify intent
node architecture/workflow-router.js "deploy the app to production"

# Validate a response
node architecture/validator.js deploy /tmp/response.md

# Enforce format
node architecture/format-locker.js coding /tmp/response.md

# Full pipeline
node architecture/orchestrator.js "your request here" --verbose

Current Tasks

• Define team agent structure
• Create agent instruction files
• Establish workflow protocols
• Test first delegation cycle
• Enhance site survey application with dashboard
• Implement memory system for context persistence
• Deploy super-enhanced memory system with JavaScript engine
• Implement workflow router with intent classification
• Deploy validation layer with safety checks
• Deploy format locker with auto-fix
• Integrate all components via orchestrator
• Implement cross-project pattern registry for retrieval-augmented generalization
• Implement feedback loop (scripts/utils/propose-pattern.js) so new workarounds can propose pattern updates

Decisions Log

Date	Decision	Rationale
2026-08-06	Create a centralized patterns/ registry and wire it into architecture/pipeline.js	Captures recurring technical lessons across projects; enables agents to generalize known solutions to new tasks as the first concrete component of a continual-learning layer.
2026-08-06	Add scripts/utils/propose-pattern.js with dry-run default and --apply gate	Agents need a lightweight, auditable way to grow the registry from real fixes without silently rewriting shared knowledge.
2026-07-03	Create dedicated dev team	Need scalable capacity for multiple internal app projects
2026-07-03	Python/FastAPI + HTMX stack	Matches existing skills, FastAPI's type safety, HTMX keeps frontend simple