fix: harden provider recommendation safety
This commit is contained in:
14
PLAYBOOK.md
14
PLAYBOOK.md
@@ -183,10 +183,10 @@ Fix:
|
||||
bun run profile:init -- --provider ollama --model llama3.1:8b
|
||||
```
|
||||
|
||||
Or auto-pick a local profile:
|
||||
Or pick a local Ollama profile automatically by goal:
|
||||
|
||||
```powershell
|
||||
bun run profile:auto -- --goal balanced
|
||||
bun run profile:init -- --provider ollama --goal balanced
|
||||
```
|
||||
|
||||
## 6.5 Placeholder key (`SUA_CHAVE`) error
|
||||
@@ -220,14 +220,16 @@ bun run profile:fast # llama3.2:3b
|
||||
bun run profile:code # qwen2.5-coder:7b
|
||||
```
|
||||
|
||||
Goal-based auto-selection:
|
||||
Goal-based local auto-selection:
|
||||
|
||||
```powershell
|
||||
bun run profile:auto -- --goal latency
|
||||
bun run profile:auto -- --goal balanced
|
||||
bun run profile:auto -- --goal coding
|
||||
bun run profile:init -- --provider ollama --goal latency
|
||||
bun run profile:init -- --provider ollama --goal balanced
|
||||
bun run profile:init -- --provider ollama --goal coding
|
||||
```
|
||||
|
||||
`profile:auto` is a best-available provider picker, not a local-only command. Use `--provider ollama` when you want to stay on a local model.
|
||||
|
||||
## 8. Practical Prompt Playbook (Copy/Paste)
|
||||
|
||||
## 8.1 Code understanding
|
||||
|
||||
Reference in New Issue
Block a user