Files
Brain/PIPELINE-STANDARD.md
ParzivalTD 06661525f8 Deploy: TekDek Command Center (2026-04-13)
- Complete Node.js + PostgreSQL application
- 10 REST API endpoints (CRUD for projects/tasks)
- Responsive HTML/CSS/JavaScript UI
- Production-ready code (95%+ test coverage)
- Deployed to /publish/web1/public/command-center/
- Server running on port 3000

Pipeline: Daedalus (arch) → Talos (code) → Icarus (UI) → Hephaestus (deploy)
Total time: 30 minutes
Token efficiency: ~783k tokens (~$6.65)

Documentation: DEPLOYMENT-POSTMORTEM-2026-04-13.md
2026-04-13 12:50:40 -04:00

144 lines
3.5 KiB
Markdown

# TekDek Development Pipeline — Standard Process
**Effective**: 2026-04-13
**Owner**: ParzivalTD
**Status**: Active
---
## Pipeline: Requirement → Live
```
Requirement → Daedalus (Arch) → Talos (Code) → Icarus (UI) → Hephaestus (Deploy) → Live
```
---
## DAEDALUS OUTPUT STANDARD (Mandatory)
Daedalus must deliver **structured JSON + checklists**, not just prose specs.
### Deliverable Format
**1. Data Model (JSON Schema)**
```json
{
"tables": [
{
"name": "table_name",
"fields": [
{"name": "id", "type": "INT PRIMARY KEY AUTO_INCREMENT"},
{"name": "field", "type": "VARCHAR(255) NOT NULL"}
],
"indexes": ["field1", "field2"],
"constraints": ["FOREIGN KEY..."]
}
]
}
```
**2. API Endpoints (Structured List)**
```json
{
"endpoints": [
{
"method": "POST",
"path": "/api/resource",
"request": {"field": "type"},
"response": {"id": "INT", "field": "type"},
"errors": [{"code": 400, "message": "..."}]
}
]
}
```
**3. Implementation Checklist (Numbered)**
```
Implementation Tasks for Talos:
1. Create database schema (copy JSON schema above)
2. Implement POST /api/resource endpoint
3. Add validation for all fields
4. Add error handling for all cases
... (numbered list, not prose)
```
**4. Architecture Notes (Prose, brief)**
- Why these design choices
- Trade-offs considered
- Assumptions made
- Performance notes
### Why This Format
- ✅ Talos parses JSON faster than prose
- ✅ Checklist = clear implementation steps (no interpretation needed)
- ✅ Reduces Talos's token spend by ~15-20%
- ✅ Fewer bugs from misinterpretation
- ✅ Pipeline savings: ~10% per cycle
---
## TALOS OUTPUT STANDARD (Mandatory)
Talos must deliver:
1. **Working code** (PHP/MySQL)
2. **Database migrations** (SQL scripts)
3. **API documentation** (with curl examples)
4. **Test coverage** (90%+ passing)
5. **Implementation checklist** (what was built, what remains)
---
## ICARUS OUTPUT STANDARD (Mandatory)
Icarus must deliver:
1. **HTML/CSS/JavaScript** (semantic, responsive)
2. **Integration guide** (how to connect to APIs)
3. **Accessibility verification** (WCAG 2.1 AA)
4. **Performance report** (Lighthouse score)
5. **Deployment checklist** (ready for Hephaestus)
---
## HEPHAESTUS OUTPUT STANDARD (Mandatory)
Hephaestus must deliver:
1. **Deployment completed** (live on web.tekdek.dev)
2. **Verification report** (all systems tested)
3. **Monitoring setup** (logging, alerts configured)
4. **Runbook** (how to maintain, troubleshoot)
5. **Go-live confirmation** (users can access)
---
## Token Efficiency Rules
1. **Daedalus output structured** (JSON + checklist)
2. **No unnecessary prose** (structured > prose)
3. **Each agent receives only what they need** (no bloat)
4. **Each agent outputs structured + documentation** (not raw code dumps)
5. **Costs tracked per cycle** (optimize monthly)
---
## Escalation Path
- Issue with spec → Ask Daedalus for clarification (structured query)
- Issue with code → Ask Talos for fix (specific endpoint/function)
- Issue with UI → Ask Icarus for adjustment (specific component)
- Issue with deployment → Ask Hephaestus for rollback/fix
---
## Success Metrics
- ✅ Pipeline time: < 24 hours from requirement to live
- ✅ Token efficiency: < 550k tokens per cycle
- ✅ Code quality: 90%+ test coverage, Lighthouse 90+
- ✅ Zero bugs in production (first 7 days)
- ✅ Full documentation delivered
---
**This is the standard. Every project follows this format.**