create-bk-agent 0.19.0 → 0.20.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +34 -1
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -11,7 +11,7 @@ npx create-bk-agent my-project
|
|
|
11
11
|
The CLI asks three questions and generates a production-ready project:
|
|
12
12
|
|
|
13
13
|
```
|
|
14
|
-
✨ create-bk-agent v0.
|
|
14
|
+
✨ create-bk-agent v0.19.1
|
|
15
15
|
|
|
16
16
|
? Project name: my-project
|
|
17
17
|
? Architecture: Multi-agent (orchestrator + specialists)
|
|
@@ -54,6 +54,39 @@ Available for `multi-agent` and `single` architectures:
|
|
|
54
54
|
|
|
55
55
|
Anthropic (Claude), OpenAI, DeepSeek / any OpenAI-compatible endpoint, and Ollama (local).
|
|
56
56
|
|
|
57
|
+
## Model configuration
|
|
58
|
+
|
|
59
|
+
Generated projects read three env vars for model selection:
|
|
60
|
+
|
|
61
|
+
| Variable | Default | Description |
|
|
62
|
+
|---|---|---|
|
|
63
|
+
| `LLM_MODEL` | `gpt-4o` | Base model — used by all agents unless overridden |
|
|
64
|
+
| `LLM_MODEL_FAST` | `LLM_MODEL` | Execution agents (specialists, tool-calling loops) |
|
|
65
|
+
| `LLM_MODEL_SMART` | `LLM_MODEL` | Routing agents (orchestrator, reasoning-heavy tasks) |
|
|
66
|
+
|
|
67
|
+
If `LLM_MODEL_FAST` and `LLM_MODEL_SMART` are not set, all agents use `LLM_MODEL` — fully backwards compatible.
|
|
68
|
+
|
|
69
|
+
Example `.env` for DeepSeek with smart/fast split:
|
|
70
|
+
|
|
71
|
+
```env
|
|
72
|
+
LLM_API_KEY=sk-...
|
|
73
|
+
LLM_BASE_URL=https://api.deepseek.com/v1
|
|
74
|
+
LLM_MODEL=deepseek-chat
|
|
75
|
+
LLM_MODEL_FAST=deepseek-chat
|
|
76
|
+
LLM_MODEL_SMART=deepseek-reasoner
|
|
77
|
+
```
|
|
78
|
+
|
|
79
|
+
The split is wired in `src/engine.ts` — specialists get `llmModelFast`, orchestrator gets `llmModelSmart`. To assign a model directly on a profile instead of via env vars, uncomment the `model:` field in the agent file:
|
|
80
|
+
|
|
81
|
+
```ts
|
|
82
|
+
// src/agents/orchestrator.ts
|
|
83
|
+
export const ORCHESTRATOR_PROFILE: AgentProfile = {
|
|
84
|
+
id: 'orchestrator',
|
|
85
|
+
model: 'deepseek-reasoner', // uncomment and set explicitly
|
|
86
|
+
...
|
|
87
|
+
};
|
|
88
|
+
```
|
|
89
|
+
|
|
57
90
|
---
|
|
58
91
|
|
|
59
92
|
## Add capabilities to an existing project
|