71dee23d3b
- Source of truth: .kilo/agents/*.md frontmatter (NOT kilo.json) - No config files in root (breaks Kilo Code) - Sync script: scripts/sync-agents.cjs --fix - Lists prohibited actions and model reference
125 lines
3.3 KiB
Markdown
125 lines
3.3 KiB
Markdown
# Evolutionary Mode Rules
|
|
|
|
When agents are modified, created, or updated during evolutionary improvement, this rule ensures all related files stay synchronized.
|
|
|
|
## Source of Truth
|
|
|
|
**`.kilo/agents/*.md` frontmatter** is the single source of truth for:
|
|
- Agent definitions (models, modes, descriptions)
|
|
- Permissions and tool access
|
|
|
|
**DO NOT create config files in root directory** - this breaks Kilo Code.
|
|
|
|
## Files to Synchronize
|
|
|
|
When agents change, update ALL of these files:
|
|
|
|
| File | What to Update |
|
|
|------|----------------|
|
|
| `.kilo/agents/agent-name.md` | Model in YAML frontmatter (source of truth) |
|
|
| `.kilo/KILO_SPEC.md` | Pipeline Agents table, Workflow Commands table |
|
|
| `AGENTS.md` | Pipeline Agents tables by category |
|
|
| `.kilo/agents/orchestrator.md` | Task Tool Invocation table (if new subagent) |
|
|
|
|
## Sync Checklist
|
|
|
|
When modifying agents:
|
|
|
|
```
|
|
□ Update agent .md file frontmatter (model, description, mode)
|
|
□ Run: node scripts/sync-agents.cjs --fix
|
|
□ Verify KILO_SPEC.md and AGENTS.md updated
|
|
□ Update orchestrator.md subagent_type mappings (if new agent)
|
|
□ Commit all changes together
|
|
```
|
|
|
|
## Adding New Agent
|
|
|
|
1. Create `.kilo/agents/agent-name.md` with frontmatter:
|
|
```yaml
|
|
---
|
|
description: Agent description
|
|
mode: subagent|primary|all
|
|
model: provider/model-id
|
|
color: #HEX
|
|
permission:
|
|
read: allow
|
|
edit: allow
|
|
...
|
|
---
|
|
```
|
|
|
|
2. If subagent, add to `orchestrator.md`:
|
|
- Add to permission list under `task:`
|
|
- Add to Task Tool Invocation table
|
|
|
|
3. Run sync:
|
|
```bash
|
|
node scripts/sync-agents.cjs --fix
|
|
```
|
|
|
|
4. Commit all changes:
|
|
```bash
|
|
git add .kilo/agents/new-agent.md .kilo/KILO_SPEC.md AGENTS.md .kilo/agents/orchestrator.md
|
|
git commit -m "feat: add new-agent agent"
|
|
```
|
|
|
|
## Model Changes
|
|
|
|
When changing a model:
|
|
|
|
1. Update `.kilo/agents/agent-name.md` frontmatter `model:` field
|
|
2. Run `node scripts/sync-agents.cjs --fix`
|
|
3. Document reason in commit message
|
|
|
|
Example:
|
|
```
|
|
fix: update LeadDeveloper model to qwen3-coder:480b
|
|
|
|
Reason: Better code generation quality, supports larger context
|
|
```
|
|
|
|
## Available Models
|
|
|
|
Use exact model IDs from `.kilo/KILO_SPEC.md` Model Format section:
|
|
|
|
| Model ID | Use Case |
|
|
|----------|----------|
|
|
| `ollama-cloud/kimi-k2-thinking` | Complex reasoning, requirements |
|
|
| `ollama-cloud/qwen3-coder:480b` | Code generation |
|
|
| `ollama-cloud/deepseek-v3.2` | Backend, DevOps |
|
|
| `ollama-cloud/minimax-m2.5` | Code review, fixing |
|
|
| `ollama-cloud/nemotron-3-super` | Performance, evaluation |
|
|
| `ollama-cloud/nemotron-3-nano:30b` | Lightweight tasks |
|
|
| `ollama-cloud/glm-5` | Orchestration |
|
|
| `ollama-cloud/gpt-oss:120b` | Large context tasks |
|
|
| `qwen/qwen3.6-plus:free` | Planning (free tier) |
|
|
|
|
## Verification
|
|
|
|
Run sync verification before commits:
|
|
|
|
```bash
|
|
# Check only (CI mode)
|
|
node scripts/sync-agents.cjs --check
|
|
|
|
# Fix discrepancies
|
|
node scripts/sync-agents.cjs --fix
|
|
```
|
|
|
|
## CI Integration
|
|
|
|
Add to `.github/workflows/ci.yml`:
|
|
|
|
```yaml
|
|
- name: Verify Agent Sync
|
|
run: node scripts/sync-agents.cjs --check
|
|
```
|
|
|
|
## Prohibited Actions
|
|
|
|
- DO NOT create `kilo.json` or `kilo-meta.json` in root (breaks Kilo Code)
|
|
- DO NOT update KILO_SPEC.md without updating agent files first
|
|
- DO NOT add new agent without updating orchestrator permissions
|
|
- DO NOT skip running sync script after changes
|
|
- DO NOT commit only partial files (always commit all sync targets) |