LLM Providers

cosh-core connects to multiple LLM providers via the OpenAI-compatible API protocol. All providers use streaming SSE output and support function calling.

Supported Providers

Provider Type	Profile	Description
dashscope	DashScope	Alibaba Cloud Bailian (Qwen series), supports thinking
aliyun	SysOM	Alibaba Cloud AK/SK signature authentication (ROA style)
openai	OpenAI	OpenAI official API, uses max_completion_tokens
deepseek	DeepSeek	DeepSeek API, supports thinking
Other	Generic	Any OpenAI-compatible endpoint

Configuration

Configure providers in ~/.copilot-shell/config.toml:

[ai]
active_model = "qwen-plus"

[ai.providers.dashscope]
type = "dashscope"
base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1"
api_key = ""    # Or via DASHSCOPE_API_KEY environment variable
model = "qwen-plus"

[ai.providers.aliyun]
type = "aliyun"
access_key_id = ""      # Or ALIBABA_CLOUD_ACCESS_KEY_ID
access_key_secret = ""  # Or ALIBABA_CLOUD_ACCESS_KEY_SECRET
model = "qwen-plus"

[ai.providers.openai]
type = "openai"
base_url = "https://api.openai.com/v1"
api_key = ""    # Or OPENAI_API_KEY
model = "gpt-4o"

[ai.providers.deepseek]
type = "deepseek"
base_url = "https://api.deepseek.com/v1"
api_key = ""
model = "deepseek-chat"

Provider Profile Differences

Behavioral differences across profiles in API requests:

Profile	max_tokens Field	Thinking Field	Authentication
Generic	max_tokens	—	Bearer token
DashScope	max_tokens	reasoning_content	Bearer token
OpenAI	max_completion_tokens	—	Bearer token
DeepSeek	max_tokens	reasoning_content	Bearer token
SysOM (aliyun)	—	—	AK/SK signature

Runtime Switching

Dynamically switch models via JSONL control protocol:

{"type":"control_request","request_id":"sw-1","request":{"subtype":"switch_model","model":"qwen-max"}}

Or override via CLI arguments:

cosh-core --headless --model qwen-max

Environment variable override:

COSH_MODEL=qwen-max cosh-core --headless

Authentication Priority

1. CLI --model argument → selects corresponding provider config
2. Environment variables (DASHSCOPE_API_KEY, ALIBABA_CLOUD_ACCESS_KEY_*)
3. Values in config.toml [ai.providers.<name>]
4. All empty → triggers interactive authentication flow (sends auth_required to Shell)

Generation Parameters

Parameter	Default	Description
max_tokens	4096	Maximum generated tokens
temperature	—	Sampling temperature (uses provider default if not set)
stream	true	Always stream output

Custom parameters can be passed via extra_params in the [ai.providers.<name>] section of config.toml.