mirror of
https://github.com/rennf93/roboco.git
synced 2026-08-03 07:23:24 +02:00
Renn F
5315e9c72d
Task Management:
- Add cancellation safeguards: require valid reason category (duplicate,
obsolete, blocked_permanently, reassigned, scope_change, stakeholder_request)
- Protect active work from arbitrary cancellation - must pause/block first
- Auto-notify PM when task is blocked with ACTION REQUIRED message
- PM task scan now shows blocked tasks needing their attention
Permissions:
- Add VIEW_STATS to Developer, QA, Documenter, Head Marketing KB permissions
- Aligns code with docs/workflows/PERMISSIONS.md specification
RAG/Embeddings:
- Upgrade embedding model from all-MiniLM-L6-v2 to nomic-embed-text-v1.5
- 768 dimensions with 8K token context (vs 512 tokens)
- Add per-index chunk sizes: docs=1536, journals=1024, others=512
- Switch to fixed chunking (semantic chunking loads separate MiniLM model)
- Add einops dependency required by nomic model
Workflow Documentation
Quick Start
| I am a... | Start here |
|---|---|
| Developer | DEVELOPER.md → AGENT_CHEATSHEET.md |
| QA | QA.md → AGENT_CHEATSHEET.md |
| Documenter | DOCUMENTER.md → AGENT_CHEATSHEET.md |
| Cell PM | PM.md → PERMISSIONS.md |
| Main PM | PM.md → PERMISSIONS.md |
Documentation Index
Core Workflows
| Document | Description |
|---|---|
| STATUS_TRANSITIONS.md | Complete task lifecycle diagram |
| PM.md | Main PM and Cell PM workflows |
| DEVELOPER.md | Developer workflow |
| QA.md | QA workflow |
| DOCUMENTER.md | Documenter workflow |
Reference
| Document | Description |
|---|---|
| PERMISSIONS.md | Tool, channel, notification permissions |
| AGENT_CHEATSHEET.md | Quick reference per role |
Activities
| Document | Description |
|---|---|
| JOURNALING.md | How to journal effectively |
| COMMUNICATION.md | Messages and channels |
| ESCALATION.md | When and how to escalate |
| KNOWLEDGE_BASE.md | Searching past work |
| GIT_WORKFLOW.md | Git conventions (future) |
Bug Tracking
| Document | Description |
|---|---|
| BUGS.md | Known issues and fixes |
The Big Picture
┌─────────────────────────────────────────────────────────────────────────────┐
│ ROBOCO WORKFLOW │
└─────────────────────────────────────────────────────────────────────────────┘
BOARD/CEO
│
│ Creates initiative
▼
MAIN PM
│
┌───────────────┼───────────────┐
│ │ │
▼ ▼ ▼
BE-PM FE-PM UX-PM
│ │ │
┌──────────┼──────────┐ │ ┌──────────┼──────────┐
│ │ │ │ │ │ │
▼ ▼ ▼ │ ▼ ▼ ▼
BE-DEV-1 BE-DEV-2 BE-QA │ FE-DEV-1 FE-DEV-2 FE-QA
│ │ │ │ │ │ │
└────┬─────┘ │ │ └────┬─────┘ │
│ │ │ │ │
▼ ▼ │ ▼ ▼
SUBMITS TO QA ───► REVIEWS │ SUBMITS TO QA ───► REVIEWS
│ │ │ │ │
▼ ▼ │ ▼ ▼
BE-DOC ◄───── QA PASSES │ FE-DOC ◄───── QA PASSES
│ │ │
▼ │ ▼
AWAITING_PM_REVIEW ◄───────┼─── AWAITING_PM_REVIEW
│ │ │
└─────────────────────┴─────────┘
│
▼
COMPLETED
Task Lifecycle Summary
BACKLOG → PENDING → CLAIMED → IN_PROGRESS → VERIFYING → AWAITING_QA
│
┌─────────────────────────┴─────────────────────────┐
│ │
QA PASSES QA FAILS
│ │
▼ ▼
AWAITING_DOCUMENTATION NEEDS_REVISION
│ │
DOCS COMPLETE (back to dev)
│
▼
AWAITING_PM_REVIEW
│
PM COMPLETES
│
▼
COMPLETED
Key Principles
- Everything is a task - All work is tracked
- Tasks start in BACKLOG - PM setup phase
- ACTIVATE before claim - Makes task visible to workers
- CLAIM before work - Takes ownership
- PLAN before START - Required planning step
- PROGRESS updates - Keep PM informed
- SELF-VERIFY first - Check your work before QA
- JOURNAL as you go - Document decisions, learnings, struggles
- No self-review - QA/Docs can't review own work
- Only PM completes - After full workflow
Common Patterns
Starting Work
# 1. Check notifications
roboco_notify_list()
roboco_notify_ack(notification_id)
# 2. Scan for tasks
roboco_task_scan(team="backend")
# 3. Search knowledge base
roboco_kb_search("similar work") # Semantic search
roboco_rag_query("how does X work?") # AI-generated answer
roboco_journal_search("past decisions") # Your journal
# 4. Claim and plan
roboco_task_claim(task_id)
roboco_task_plan(task_id, approach, steps)
# 5. Start
roboco_task_start(task_id)
While Working
# Progress updates
roboco_task_progress(task_id, "Completed X", 50)
# Journaling
roboco_journal_decision({...})
roboco_journal_learning({...})
# Communication
roboco_message_send({channel: "backend-cell", ...})
# If stuck
roboco_task_escalate(task_id, "Need help with X")
# If you can't continue (graceful exit)
roboco_task_substitute(task_id, "low_context", "Need more context about X")
Finishing
# Self-verify
roboco_task_submit_verification(task_id)
# Submit for QA
roboco_task_submit_qa(task_id, notes)
# Reflect
roboco_journal_reflect({...})