dgf1988/codewhale
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

feat(provider): add Ollama provider support (#921)

Browse Source 
Source PR: #921 by @reidliu41.
Closes #908.

Local verification:
- cargo test --workspace --all-features ollama
- cargo fmt --all -- --check
- cargo build

Co-authored-by: reidliu41 <reid201711@gmail.com>
This commit is contained in:
Reid
2026-05-07 09:16:46 +08:00
committed by GitHub
parent 8a0a166b9c
commit 78c415f40c
17 changed files with 581 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+8 -2
View File
@@ -197,6 +197,10 @@ SGLANG_BASE_URL="http://localhost:30000/v1" deepseek --provider sglang --model d
# Self-hosted vLLM
VLLM_BASE_URL="http://localhost:8000/v1" deepseek --provider vllm --model deepseek-v4-flash
# Self-hosted Ollama
ollama pull deepseek-coder:1.3b
deepseek --provider ollama --model deepseek-coder:1.3b
```
---
@@ -306,12 +310,14 @@ Key environment variables:
| `DEEPSEEK_BASE_URL` | API base URL |
| `DEEPSEEK_HTTP_HEADERS` | Optional custom model request headers, e.g. `X-Model-Provider-Id=your-model-provider` |
| `DEEPSEEK_MODEL` | Default model |
| `DEEPSEEK_PROVIDER` | `deepseek` (default), `nvidia-nim`, `fireworks`, `sglang`, `vllm` |
| `DEEPSEEK_PROVIDER` | `deepseek` (default), `nvidia-nim`, `fireworks`, `sglang`, `vllm`, `ollama` |
| `DEEPSEEK_PROFILE` | Config profile name |
| `DEEPSEEK_MEMORY` | Set to `on` to enable user memory |
| `NVIDIA_API_KEY` / `FIREWORKS_API_KEY` / `SGLANG_API_KEY` / `VLLM_API_KEY` | Provider auth |
| `NVIDIA_API_KEY` / `FIREWORKS_API_KEY` / `SGLANG_API_KEY` / `VLLM_API_KEY` / `OLLAMA_API_KEY` | Provider auth |
| `SGLANG_BASE_URL` | Self-hosted SGLang endpoint |
| `VLLM_BASE_URL` | Self-hosted vLLM endpoint |
| `OLLAMA_BASE_URL` | Self-hosted Ollama endpoint |
| `OLLAMA_MODEL` | Self-hosted Ollama model tag |
| `NO_ANIMATIONS=1` | Force accessibility mode at startup |
| `SSL_CERT_FILE` | Custom CA bundle for corporate proxies |
README.zh-CN.md
+8 -2
View File
@@ -172,6 +172,10 @@ SGLANG_BASE_URL="http://localhost:30000/v1" deepseek --provider sglang --model d
# 自托管 vLLM
VLLM_BASE_URL="http://localhost:8000/v1" deepseek --provider vllm --model deepseek-v4-flash
# 自托管 Ollama
ollama pull deepseek-coder:1.3b
deepseek --provider ollama --model deepseek-coder:1.3b
```
---
@@ -256,12 +260,14 @@ deepseek update # 检查并应用二进制更新
| `DEEPSEEK_API_KEY` | DeepSeek API key |
| `DEEPSEEK_BASE_URL` | API base URL |
| `DEEPSEEK_MODEL` | 默认模型 |
| `DEEPSEEK_PROVIDER` | `deepseek`（默认）、`nvidia-nim`、`fireworks`、`sglang`、`vllm` |
| `DEEPSEEK_PROVIDER` | `deepseek`（默认）、`nvidia-nim`、`fireworks`、`sglang`、`vllm`、`ollama` |
| `DEEPSEEK_PROFILE` | 配置 profile 名称 |
| `DEEPSEEK_MEMORY` | 设为 `on` 启用用户记忆 |
| `NVIDIA_API_KEY` / `FIREWORKS_API_KEY` / `SGLANG_API_KEY` / `VLLM_API_KEY` | 提供商认证 |
| `NVIDIA_API_KEY` / `FIREWORKS_API_KEY` / `SGLANG_API_KEY` / `VLLM_API_KEY` / `OLLAMA_API_KEY` | 提供商认证 |
| `SGLANG_BASE_URL` | 自托管 SGLang 端点 |
| `VLLM_BASE_URL` | 自托管 vLLM 端点 |
| `OLLAMA_BASE_URL` | 自托管 Ollama 端点 |
| `OLLAMA_MODEL` | 自托管 Ollama 模型标签 |
| `NO_ANIMATIONS=1` | 启动时强制无障碍模式 |
| `SSL_CERT_FILE` | 企业代理的自定义 CA 包 |
config.example.toml
+9 -2
View File
@@ -12,10 +12,10 @@
# Choose which provider to use by default. Per-provider credentials live in the
# `[providers.*]` sections near the bottom of
# this file — keeping both stored at once means `/provider deepseek` and
# `/provider nvidia-nim` (or `--provider fireworks`, `/provider sglang`, `/provider vllm`) toggle without having to
# `/provider nvidia-nim` (or `--provider fireworks`, `/provider sglang`, `/provider vllm`, `/provider ollama`) toggle without having to
# re-enter keys. Top-level `api_key` / `base_url` are still read as DeepSeek
# defaults when `[providers.deepseek]` is absent (backward compatibility).
provider = "deepseek" # deepseek | nvidia-nim | openrouter | novita | fireworks | sglang | vllm
provider = "deepseek" # deepseek | nvidia-nim | openrouter | novita | fireworks | sglang | vllm | ollama
api_key = "YOUR_DEEPSEEK_API_KEY" # must be non-empty
base_url = "https://api.deepseek.com"
# base_url = "https://api.deepseeki.com" # China users
@@ -158,6 +158,7 @@ max_subagents = 10 # optional (1-20)
# Fireworks: FIREWORKS_API_KEY, FIREWORKS_BASE_URL
# SGLang: SGLANG_BASE_URL, SGLANG_MODEL, optional SGLANG_API_KEY
# vLLM: VLLM_BASE_URL, VLLM_MODEL, optional VLLM_API_KEY
# Ollama: OLLAMA_BASE_URL, OLLAMA_MODEL, optional OLLAMA_API_KEY
# DeepSeek Platform (https://platform.deepseek.com)
[providers.deepseek]
@@ -190,6 +191,12 @@ max_subagents = 10 # optional (1-20)
# base_url = "http://localhost:8000/v1"
# model = "deepseek-ai/DeepSeek-V4-Pro" # or deepseek-ai/DeepSeek-V4-Flash
# Self-hosted Ollama OpenAI-compatible server
[providers.ollama]
# api_key = "OPTIONAL_OLLAMA_TOKEN"
# base_url = "http://localhost:11434/v1"
# model = "deepseek-coder:1.3b" # or any local Ollama tag
# ─────────────────────────────────────────────────────────────────────────────────
# Network Policy (#135)
# ─────────────────────────────────────────────────────────────────────────────────
crates/agent/src/lib.rs
+41
View File
@@ -185,6 +185,13 @@ impl Default for ModelRegistry {
supports_tools: true,
supports_reasoning: true,
},
ModelInfo {
id: "deepseek-coder:1.3b".to_string(),
provider: ProviderKind::Ollama,
aliases: vec![],
supports_tools: true,
supports_reasoning: false,
},
];
Self::new(models)
}
@@ -218,6 +225,20 @@ impl ModelRegistry {
if let Some(name) = requested {
fallback_chain.push(format!("requested:{name}"));
if provider_hint == Some(ProviderKind::Ollama) {
return ModelResolution {
requested: Some(name.to_string()),
resolved: ModelInfo {
id: name.trim().to_string(),
provider: ProviderKind::Ollama,
aliases: Vec::new(),
supports_tools: true,
supports_reasoning: false,
},
used_fallback: false,
fallback_chain,
};
}
if let Some(provider) = provider_hint
&& let Some(model) = self
.models
@@ -406,6 +427,26 @@ mod tests {
assert_eq!(resolved.resolved.id, "deepseek-ai/DeepSeek-V4-Pro");
}
#[test]
fn ollama_default_uses_small_local_model_id() {
let registry = ModelRegistry::default();
let resolved = registry.resolve(None, Some(ProviderKind::Ollama));
assert_eq!(resolved.resolved.provider, ProviderKind::Ollama);
assert_eq!(resolved.resolved.id, "deepseek-coder:1.3b");
assert!(!resolved.resolved.supports_reasoning);
}
#[test]
fn ollama_requested_model_tag_is_preserved() {
let registry = ModelRegistry::default();
let resolved = registry.resolve(Some("qwen2.5-coder:7b"), Some(ProviderKind::Ollama));
assert_eq!(resolved.resolved.provider, ProviderKind::Ollama);
assert_eq!(resolved.resolved.id, "qwen2.5-coder:7b");
assert!(!resolved.used_fallback);
}
#[test]
fn deepseek_v4_flash_alias_resolves_to_vllm_when_provider_hinted() {
let registry = ModelRegistry::default();
crates/cli/src/lib.rs
+63 -2
View File
@@ -31,6 +31,7 @@ enum ProviderArg {
Fireworks,
Sglang,
Vllm,
Ollama,
}
impl From<ProviderArg> for ProviderKind {
@@ -44,6 +45,7 @@ impl From<ProviderArg> for ProviderKind {
ProviderArg::Fireworks => ProviderKind::Fireworks,
ProviderArg::Sglang => ProviderKind::Sglang,
ProviderArg::Vllm => ProviderKind::Vllm,
ProviderArg::Ollama => ProviderKind::Ollama,
}
}
}
@@ -663,11 +665,12 @@ fn provider_slot(provider: ProviderKind) -> &'static str {
ProviderKind::Fireworks => "fireworks",
ProviderKind::Sglang => "sglang",
ProviderKind::Vllm => "vllm",
ProviderKind::Ollama => "ollama",
}
}
/// Provider order used by the `auth list` and `auth status` outputs.
const PROVIDER_LIST: [ProviderKind; 8] = [
const PROVIDER_LIST: [ProviderKind; 9] = [
ProviderKind::Deepseek,
ProviderKind::NvidiaNim,
ProviderKind::Openrouter,
@@ -675,6 +678,7 @@ const PROVIDER_LIST: [ProviderKind; 8] = [
ProviderKind::Fireworks,
ProviderKind::Sglang,
ProviderKind::Vllm,
ProviderKind::Ollama,
ProviderKind::Openai,
];
@@ -795,6 +799,19 @@ fn run_auth_command_with_secrets(
} => {
let provider: ProviderKind = provider.into();
let slot = provider_slot(provider);
if provider == ProviderKind::Ollama && api_key.is_none() && !api_key_stdin {
store.config.provider = provider;
let provider_cfg = store.config.providers.for_provider_mut(provider);
if provider_cfg.base_url.is_none() {
provider_cfg.base_url = Some("http://localhost:11434/v1".to_string());
}
store.save()?;
println!(
"configured {slot} provider in {} (API key optional)",
store.path().display()
);
return Ok(());
}
let api_key = match (api_key, api_key_stdin) {
(Some(v), _) => v,
(None, true) => read_api_key_from_stdin()?,
@@ -1284,9 +1301,10 @@ fn build_tui_command(
| ProviderKind::Fireworks
| ProviderKind::Sglang
| ProviderKind::Vllm
| ProviderKind::Ollama
) {
bail!(
"The interactive TUI supports DeepSeek, NVIDIA NIM, OpenRouter, Novita, Fireworks, SGLang, and vLLM providers. Remove --provider {} or use `deepseek model ...` for provider registry inspection.",
"The interactive TUI supports DeepSeek, NVIDIA NIM, OpenRouter, Novita, Fireworks, SGLang, vLLM, and Ollama providers. Remove --provider {} or use `deepseek model ...` for provider registry inspection.",
resolved_runtime.provider.as_str()
);
}
@@ -1892,6 +1910,18 @@ mod tests {
}))
));
let cli = parse_ok(&["deepseek", "auth", "set", "--provider", "ollama"]);
assert!(matches!(
cli.command,
Some(Commands::Auth(AuthArgs {
command: AuthCommand::Set {
provider: ProviderArg::Ollama,
api_key: None,
api_key_stdin: false,
}
}))
));
let cli = parse_ok(&["deepseek", "auth", "list"]);
assert!(matches!(
cli.command,
@@ -1957,6 +1987,37 @@ mod tests {
let _ = std::fs::remove_file(path);
}
#[test]
fn auth_set_ollama_accepts_empty_key_and_records_base_url() {
let nanos = chrono::Utc::now().timestamp_nanos_opt().unwrap_or_default();
let path = std::env::temp_dir().join(format!(
"deepseek-cli-auth-ollama-test-{}-{nanos}.toml",
std::process::id()
));
let mut store = ConfigStore::load(Some(path.clone())).expect("store should load");
let secrets = no_keyring_secrets();
run_auth_command_with_secrets(
&mut store,
AuthCommand::Set {
provider: ProviderArg::Ollama,
api_key: None,
api_key_stdin: false,
},
&secrets,
)
.expect("ollama auth set should not require a key");
assert_eq!(store.config.provider, ProviderKind::Ollama);
assert_eq!(
store.config.providers.ollama.base_url.as_deref(),
Some("http://localhost:11434/v1")
);
assert_eq!(store.config.providers.ollama.api_key, None);
let _ = std::fs::remove_file(path);
}
#[test]
fn auth_clear_removes_from_config() {
use deepseek_secrets::{InMemoryKeyringStore, KeyringStore};
crates/config/src/lib.rs
+120
View File
@@ -35,6 +35,8 @@ const DEFAULT_SGLANG_BASE_URL: &str = "http://localhost:30000/v1";
const DEFAULT_VLLM_MODEL: &str = "deepseek-ai/DeepSeek-V4-Pro";
const DEFAULT_VLLM_FLASH_MODEL: &str = "deepseek-ai/DeepSeek-V4-Flash";
const DEFAULT_VLLM_BASE_URL: &str = "http://localhost:8000/v1";
const DEFAULT_OLLAMA_MODEL: &str = "deepseek-coder:1.3b";
const DEFAULT_OLLAMA_BASE_URL: &str = "http://localhost:11434/v1";
#[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq, Default)]
#[serde(rename_all = "kebab-case")]
@@ -48,6 +50,7 @@ pub enum ProviderKind {
Fireworks,
Sglang,
Vllm,
Ollama,
}
impl ProviderKind {
@@ -62,6 +65,7 @@ impl ProviderKind {
Self::Fireworks => "fireworks",
Self::Sglang => "sglang",
Self::Vllm => "vllm",
Self::Ollama => "ollama",
}
}
@@ -76,6 +80,7 @@ impl ProviderKind {
"fireworks" | "fireworks-ai" => Some(Self::Fireworks),
"sglang" | "sg-lang" => Some(Self::Sglang),
"vllm" | "v-llm" => Some(Self::Vllm),
"ollama" | "ollama-local" => Some(Self::Ollama),
_ => None,
}
}
@@ -108,6 +113,8 @@ pub struct ProvidersToml {
pub sglang: ProviderConfigToml,
#[serde(default)]
pub vllm: ProviderConfigToml,
#[serde(default)]
pub ollama: ProviderConfigToml,
}
impl ProvidersToml {
@@ -122,6 +129,7 @@ impl ProvidersToml {
ProviderKind::Fireworks => &self.fireworks,
ProviderKind::Sglang => &self.sglang,
ProviderKind::Vllm => &self.vllm,
ProviderKind::Ollama => &self.ollama,
}
}
@@ -135,6 +143,7 @@ impl ProvidersToml {
ProviderKind::Fireworks => &mut self.fireworks,
ProviderKind::Sglang => &mut self.sglang,
ProviderKind::Vllm => &mut self.vllm,
ProviderKind::Ollama => &mut self.ollama,
}
}
}
@@ -343,6 +352,7 @@ impl ConfigToml {
merge_provider_config(&mut self.providers.fireworks, &project.providers.fireworks);
merge_provider_config(&mut self.providers.sglang, &project.providers.sglang);
merge_provider_config(&mut self.providers.vllm, &project.providers.vllm);
merge_provider_config(&mut self.providers.ollama, &project.providers.ollama);
if project.network.is_some() {
self.network = project.network;
@@ -426,6 +436,12 @@ impl ConfigToml {
"providers.vllm.http_headers" => {
serialize_http_headers(&self.providers.vllm.http_headers)
}
"providers.ollama.api_key" => self.providers.ollama.api_key.clone(),
"providers.ollama.base_url" => self.providers.ollama.base_url.clone(),
"providers.ollama.model" => self.providers.ollama.model.clone(),
"providers.ollama.http_headers" => {
serialize_http_headers(&self.providers.ollama.http_headers)
}
_ => self.extras.get(key).map(toml::Value::to_string),
}
}
@@ -549,6 +565,18 @@ impl ConfigToml {
"providers.vllm.http_headers" => {
self.providers.vllm.http_headers = parse_http_headers(value)?;
}
"providers.ollama.api_key" => {
self.providers.ollama.api_key = Some(value.to_string());
}
"providers.ollama.base_url" => {
self.providers.ollama.base_url = Some(value.to_string());
}
"providers.ollama.model" => {
self.providers.ollama.model = Some(value.to_string());
}
"providers.ollama.http_headers" => {
self.providers.ollama.http_headers = parse_http_headers(value)?;
}
_ => {
self.extras
.insert(key.to_string(), toml::Value::String(value.to_string()));
@@ -617,6 +645,10 @@ impl ConfigToml {
"providers.vllm.base_url" => self.providers.vllm.base_url = None,
"providers.vllm.model" => self.providers.vllm.model = None,
"providers.vllm.http_headers" => self.providers.vllm.http_headers.clear(),
"providers.ollama.api_key" => self.providers.ollama.api_key = None,
"providers.ollama.base_url" => self.providers.ollama.base_url = None,
"providers.ollama.model" => self.providers.ollama.model = None,
"providers.ollama.http_headers" => self.providers.ollama.http_headers.clear(),
_ => {
self.extras.remove(key);
}
@@ -764,6 +796,18 @@ impl ConfigToml {
if let Some(v) = serialize_http_headers(&self.providers.vllm.http_headers) {
out.insert("providers.vllm.http_headers".to_string(), v);
}
if let Some(v) = self.providers.ollama.api_key.as_ref() {
out.insert("providers.ollama.api_key".to_string(), redact_secret(v));
}
if let Some(v) = self.providers.ollama.base_url.as_ref() {
out.insert("providers.ollama.base_url".to_string(), v.clone());
}
if let Some(v) = self.providers.ollama.model.as_ref() {
out.insert("providers.ollama.model".to_string(), v.clone());
}
if let Some(v) = serialize_http_headers(&self.providers.ollama.http_headers) {
out.insert("providers.ollama.http_headers".to_string(), v);
}
for (k, v) in &self.extras {
out.insert(k.clone(), v.to_string());
@@ -841,6 +885,7 @@ impl ConfigToml {
ProviderKind::Fireworks => DEFAULT_FIREWORKS_BASE_URL.to_string(),
ProviderKind::Sglang => DEFAULT_SGLANG_BASE_URL.to_string(),
ProviderKind::Vllm => DEFAULT_VLLM_BASE_URL.to_string(),
ProviderKind::Ollama => DEFAULT_OLLAMA_BASE_URL.to_string(),
});
let model = cli
@@ -859,6 +904,7 @@ impl ConfigToml {
ProviderKind::Fireworks => DEFAULT_FIREWORKS_MODEL.to_string(),
ProviderKind::Sglang => DEFAULT_SGLANG_MODEL.to_string(),
ProviderKind::Vllm => DEFAULT_VLLM_MODEL.to_string(),
ProviderKind::Ollama => DEFAULT_OLLAMA_MODEL.to_string(),
});
let model = normalize_model_for_provider(provider, &model);
@@ -944,6 +990,10 @@ pub fn load_project_config(workspace: &Path) -> Option<ConfigToml> {
}
fn normalize_model_for_provider(provider: ProviderKind, model: &str) -> String {
if matches!(provider, ProviderKind::Ollama) {
return model.to_string();
}
let normalized = model.trim().to_ascii_lowercase();
match (provider, normalized.as_str()) {
(ProviderKind::NvidiaNim, "deepseek-v4-pro" | "deepseek-v4pro") => {
@@ -1222,6 +1272,7 @@ struct EnvRuntimeOverrides {
fireworks_base_url: Option<String>,
sglang_base_url: Option<String>,
vllm_base_url: Option<String>,
ollama_base_url: Option<String>,
}
impl EnvRuntimeOverrides {
@@ -1269,6 +1320,9 @@ impl EnvRuntimeOverrides {
vllm_base_url: std::env::var("VLLM_BASE_URL")
.ok()
.filter(|v| !v.trim().is_empty()),
ollama_base_url: std::env::var("OLLAMA_BASE_URL")
.ok()
.filter(|v| !v.trim().is_empty()),
}
}
@@ -1284,6 +1338,7 @@ impl EnvRuntimeOverrides {
ProviderKind::Fireworks => self.fireworks_base_url.clone(),
ProviderKind::Sglang => self.sglang_base_url.clone(),
ProviderKind::Vllm => self.vllm_base_url.clone(),
ProviderKind::Ollama => self.ollama_base_url.clone(),
}
}
}
@@ -1321,6 +1376,8 @@ mod tests {
sglang_base_url: Option<OsString>,
vllm_api_key: Option<OsString>,
vllm_base_url: Option<OsString>,
ollama_api_key: Option<OsString>,
ollama_base_url: Option<OsString>,
}
impl EnvGuard {
@@ -1346,6 +1403,8 @@ mod tests {
sglang_base_url: env::var_os("SGLANG_BASE_URL"),
vllm_api_key: env::var_os("VLLM_API_KEY"),
vllm_base_url: env::var_os("VLLM_BASE_URL"),
ollama_api_key: env::var_os("OLLAMA_API_KEY"),
ollama_base_url: env::var_os("OLLAMA_BASE_URL"),
};
// Safety: test-only environment mutation guarded by a module mutex.
unsafe {
@@ -1369,6 +1428,8 @@ mod tests {
env::remove_var("SGLANG_BASE_URL");
env::remove_var("VLLM_API_KEY");
env::remove_var("VLLM_BASE_URL");
env::remove_var("OLLAMA_API_KEY");
env::remove_var("OLLAMA_BASE_URL");
}
guard
}
@@ -1406,6 +1467,8 @@ mod tests {
Self::restore_var("SGLANG_BASE_URL", self.sglang_base_url.take());
Self::restore_var("VLLM_API_KEY", self.vllm_api_key.take());
Self::restore_var("VLLM_BASE_URL", self.vllm_base_url.take());
Self::restore_var("OLLAMA_API_KEY", self.ollama_api_key.take());
Self::restore_var("OLLAMA_BASE_URL", self.ollama_base_url.take());
}
}
}
@@ -1705,6 +1768,11 @@ mod tests {
assert_eq!(ProviderKind::parse("sg-lang"), Some(ProviderKind::Sglang));
assert_eq!(ProviderKind::parse("v-llm"), Some(ProviderKind::Vllm));
assert_eq!(ProviderKind::parse("vllm"), Some(ProviderKind::Vllm));
assert_eq!(ProviderKind::parse("ollama"), Some(ProviderKind::Ollama));
assert_eq!(
ProviderKind::parse("ollama-local"),
Some(ProviderKind::Ollama)
);
}
#[test]
@@ -1787,6 +1855,58 @@ mod tests {
assert_eq!(resolved.model, DEFAULT_VLLM_MODEL);
}
#[test]
fn ollama_provider_defaults_to_local_endpoint_and_small_model() {
let _lock = env_lock();
let _env = EnvGuard::without_deepseek_runtime_overrides();
let config = ConfigToml {
provider: ProviderKind::Ollama,
..ConfigToml::default()
};
let resolved = config.resolve_runtime_options(&CliRuntimeOverrides::default());
assert_eq!(resolved.provider, ProviderKind::Ollama);
assert_eq!(resolved.base_url, DEFAULT_OLLAMA_BASE_URL);
assert_eq!(resolved.model, DEFAULT_OLLAMA_MODEL);
assert_eq!(resolved.api_key, None);
}
#[test]
fn ollama_provider_preserves_model_tags() {
let _lock = env_lock();
let _env = EnvGuard::without_deepseek_runtime_overrides();
let cli = CliRuntimeOverrides {
provider: Some(ProviderKind::Ollama),
model: Some("deepseek-coder-v2:16b".to_string()),
..CliRuntimeOverrides::default()
};
let resolved = ConfigToml::default().resolve_runtime_options(&cli);
assert_eq!(resolved.provider, ProviderKind::Ollama);
assert_eq!(resolved.model, "deepseek-coder-v2:16b");
}
#[test]
fn ollama_env_overrides_provider_base_url_and_optional_key() {
let _lock = env_lock();
let _env = EnvGuard::without_deepseek_runtime_overrides();
// Safety: test-only environment mutation guarded by a module mutex.
unsafe {
env::set_var("DEEPSEEK_PROVIDER", "ollama-local");
env::set_var("OLLAMA_BASE_URL", "http://ollama.example/v1");
env::set_var("OLLAMA_API_KEY", "ollama-env-key");
}
let resolved =
ConfigToml::default().resolve_runtime_options(&CliRuntimeOverrides::default());
assert_eq!(resolved.provider, ProviderKind::Ollama);
assert_eq!(resolved.base_url, "http://ollama.example/v1");
assert_eq!(resolved.api_key.as_deref(), Some("ollama-env-key"));
}
#[test]
fn openrouter_env_api_key_falls_back_when_config_missing() {
let _lock = env_lock();
crates/secrets/src/lib.rs
+15
View File
@@ -435,6 +435,7 @@ pub fn env_for(name: &str) -> Option<String> {
"fireworks" | "fireworks-ai" => &["FIREWORKS_API_KEY"],
"sglang" | "sg-lang" => &["SGLANG_API_KEY"],
"vllm" | "v-llm" => &["VLLM_API_KEY"],
"ollama" | "ollama-local" => &["OLLAMA_API_KEY"],
"openai" => &["OPENAI_API_KEY"],
_ => return None,
};
@@ -472,6 +473,7 @@ mod tests {
"FIREWORKS_API_KEY",
"SGLANG_API_KEY",
"VLLM_API_KEY",
"OLLAMA_API_KEY",
"OPENAI_API_KEY",
] {
// Safety: tests serialise on env_lock(); the broader
@@ -609,6 +611,19 @@ mod tests {
unsafe { std::env::remove_var("VLLM_API_KEY") };
}
#[test]
fn ollama_env_aliases_resolve() {
let _lock = env_lock();
clear_known_envs();
// Safety: env mutation guarded by env_lock().
unsafe { std::env::set_var("OLLAMA_API_KEY", "ollama-key") };
assert_eq!(env_for("ollama").as_deref(), Some("ollama-key"));
assert_eq!(env_for("ollama-local").as_deref(), Some("ollama-key"));
// Safety: env mutation guarded by env_lock().
unsafe { std::env::remove_var("OLLAMA_API_KEY") };
}
#[cfg(unix)]
#[test]
fn file_store_round_trips_with_secure_perms() {
crates/tui/src/client.rs
+23
View File
@@ -799,6 +799,7 @@ pub(super) fn apply_reasoning_effort(
| ApiProvider::Vllm => {
body["thinking"] = json!({ "type": "disabled" });
}
ApiProvider::Ollama => {}
ApiProvider::NvidiaNim => {
body["chat_template_kwargs"] = json!({
"thinking": false,
@@ -816,6 +817,7 @@ pub(super) fn apply_reasoning_effort(
body["reasoning_effort"] = json!("high");
body["thinking"] = json!({ "type": "enabled" });
}
ApiProvider::Ollama => {}
ApiProvider::NvidiaNim => {
body["chat_template_kwargs"] = json!({
"thinking": true,
@@ -834,6 +836,7 @@ pub(super) fn apply_reasoning_effort(
body["reasoning_effort"] = json!("max");
body["thinking"] = json!({ "type": "enabled" });
}
ApiProvider::Ollama => {}
ApiProvider::NvidiaNim => {
body["chat_template_kwargs"] = json!({
"thinking": true,
@@ -1745,6 +1748,26 @@ mod tests {
);
}
#[test]
fn parse_models_response_accepts_ollama_tag_ids() {
let payload = r#"{
"object": "list",
"data": [
{"id": "qwen2.5-coder:7b", "object": "model", "owned_by": "library"},
{"id": "deepseek-coder-v2:16b", "object": "model"}
]
}"#;
let models = parse_models_response(payload).expect("parse models");
assert_eq!(
models
.iter()
.map(|model| model.id.as_str())
.collect::<Vec<_>>(),
vec!["deepseek-coder-v2:16b", "qwen2.5-coder:7b"]
);
}
#[test]
fn parse_usage_reads_deepseek_cache_and_reasoning_tokens() {
let usage = parse_usage(Some(&json!({
crates/tui/src/commands/provider.rs
+15 -1
View File
@@ -27,12 +27,13 @@ pub fn provider(app: &mut App, args: Option<&str>) -> CommandResult {
let Some(target) = ApiProvider::parse(name) else {
return CommandResult::error(format!(
"Unknown provider '{name}'. Expected: deepseek, nvidia-nim, openrouter, novita, fireworks, sglang, or vllm."
"Unknown provider '{name}'. Expected: deepseek, nvidia-nim, openrouter, novita, fireworks, sglang, vllm, or ollama."
));
};
let model = match model_arg {
None => None,
Some(raw) if target == ApiProvider::Ollama => Some(raw.trim().to_string()),
Some(raw) => match normalize_model_name(&expand_model_alias(raw)) {
Some(normalized) => Some(normalized),
None => {
@@ -180,6 +181,19 @@ mod tests {
}
}
#[test]
fn switch_to_ollama_preserves_model_tag() {
let mut app = create_test_app();
let result = provider(&mut app, Some("ollama qwen2.5-coder:7b"));
match result.action {
Some(AppAction::SwitchProvider { provider, model }) => {
assert_eq!(provider, ApiProvider::Ollama);
assert_eq!(model.as_deref(), Some("qwen2.5-coder:7b"));
}
other => panic!("expected SwitchProvider, got {other:?}"),
}
}
#[test]
fn switching_to_active_provider_without_model_is_a_noop() {
let mut app = create_test_app();
crates/tui/src/config.rs
+192 -10
View File
@@ -37,6 +37,8 @@ pub const DEFAULT_SGLANG_BASE_URL: &str = "http://localhost:30000/v1";
pub const DEFAULT_VLLM_MODEL: &str = "deepseek-ai/DeepSeek-V4-Pro";
pub const DEFAULT_VLLM_FLASH_MODEL: &str = "deepseek-ai/DeepSeek-V4-Flash";
pub const DEFAULT_VLLM_BASE_URL: &str = "http://localhost:8000/v1";
pub const DEFAULT_OLLAMA_MODEL: &str = "deepseek-coder:1.3b";
pub const DEFAULT_OLLAMA_BASE_URL: &str = "http://localhost:11434/v1";
pub const DEFAULT_DEEPSEEKCN_BASE_URL: &str = "https://api.deepseeki.com";
const API_KEYRING_SENTINEL: &str = "__KEYRING__";
pub const COMMON_DEEPSEEK_MODELS: &[&str] = &[
@@ -59,6 +61,7 @@ pub enum ApiProvider {
Fireworks,
Sglang,
Vllm,
Ollama,
}
impl ApiProvider {
@@ -75,6 +78,7 @@ impl ApiProvider {
"fireworks" | "fireworks-ai" => Some(Self::Fireworks),
"sglang" | "sg-lang" => Some(Self::Sglang),
"vllm" | "v-llm" => Some(Self::Vllm),
"ollama" | "ollama-local" => Some(Self::Ollama),
_ => None,
}
}
@@ -90,6 +94,7 @@ impl ApiProvider {
Self::Fireworks => "fireworks",
Self::Sglang => "sglang",
Self::Vllm => "vllm",
Self::Ollama => "ollama",
}
}
@@ -105,6 +110,7 @@ impl ApiProvider {
Self::Fireworks => "Fireworks AI",
Self::Sglang => "SGLang",
Self::Vllm => "vLLM",
Self::Ollama => "Ollama",
}
}
@@ -120,6 +126,7 @@ impl ApiProvider {
Self::Fireworks,
Self::Sglang,
Self::Vllm,
Self::Ollama,
]
}
}
@@ -165,6 +172,18 @@ pub enum RequestPayloadMode {
/// in the API payload (after normalization / provider-specific mapping).
#[must_use]
pub fn provider_capability(provider: ApiProvider, resolved_model: &str) -> ProviderCapability {
if matches!(provider, ApiProvider::Ollama) {
return ProviderCapability {
provider,
resolved_model: resolved_model.to_string(),
context_window: 8192,
max_output: 4096,
thinking_supported: false,
cache_telemetry_supported: false,
request_payload_mode: RequestPayloadMode::ChatCompletions,
};
}
let model_lower = resolved_model.to_ascii_lowercase();
let is_v4_pro = model_lower.contains("v4-pro") || model_lower == "deepseek-v4pro";
let is_v4_flash = model_lower.contains("v4-flash")
@@ -940,6 +959,8 @@ pub struct ProvidersConfig {
pub sglang: ProviderConfig,
#[serde(default)]
pub vllm: ProviderConfig,
#[serde(default)]
pub ollama: ProviderConfig,
}
#[derive(Debug, Clone, Deserialize, Default)]
@@ -999,7 +1020,7 @@ impl Config {
&& ApiProvider::parse(provider).is_none()
{
anyhow::bail!(
"Invalid provider '{provider}': expected deepseek, deepseek-cn, nvidia-nim, openrouter, novita, fireworks, sglang, or vllm."
"Invalid provider '{provider}': expected deepseek, deepseek-cn, nvidia-nim, openrouter, novita, fireworks, sglang, vllm, or ollama."
);
}
if let Some(ref key) = self.api_key
@@ -1016,6 +1037,7 @@ impl Config {
}
if let Some(model) = self.default_text_model.as_deref()
&& !model.trim().eq_ignore_ascii_case("auto")
&& !matches!(self.api_provider(), ApiProvider::Ollama)
&& normalize_model_name(model).is_none()
{
anyhow::bail!(
@@ -1118,6 +1140,7 @@ impl Config {
ApiProvider::Fireworks => &providers.fireworks,
ApiProvider::Sglang => &providers.sglang,
ApiProvider::Vllm => &providers.vllm,
ApiProvider::Ollama => &providers.ollama,
})
}
@@ -1144,9 +1167,18 @@ impl Config {
if let Some(model) = self
.provider_config()
.and_then(|provider| provider.model.as_deref())
&& let Some(normalized) = normalize_model_for_provider(provider, model)
{
return normalized;
if matches!(provider, ApiProvider::Ollama) {
return model.trim().to_string();
}
if let Some(normalized) = normalize_model_for_provider(provider, model) {
return normalized;
}
}
if let Some(model) = self.default_text_model.as_deref()
&& matches!(provider, ApiProvider::Ollama)
{
return model.trim().to_string();
}
if let Some(model) = self.default_text_model.as_deref()
&& model.trim().eq_ignore_ascii_case("auto")
@@ -1167,6 +1199,7 @@ impl Config {
ApiProvider::Fireworks => DEFAULT_FIREWORKS_MODEL,
ApiProvider::Sglang => DEFAULT_SGLANG_MODEL,
ApiProvider::Vllm => DEFAULT_VLLM_MODEL,
ApiProvider::Ollama => DEFAULT_OLLAMA_MODEL,
}
.to_string()
}
@@ -1193,7 +1226,8 @@ impl Config {
| ApiProvider::Novita
| ApiProvider::Fireworks
| ApiProvider::Sglang
| ApiProvider::Vllm => None,
| ApiProvider::Vllm
| ApiProvider::Ollama => None,
};
let base = provider_base.or(root_base).unwrap_or_else(|| {
match provider {
@@ -1205,6 +1239,7 @@ impl Config {
ApiProvider::Fireworks => DEFAULT_FIREWORKS_BASE_URL,
ApiProvider::Sglang => DEFAULT_SGLANG_BASE_URL,
ApiProvider::Vllm => DEFAULT_VLLM_BASE_URL,
ApiProvider::Ollama => DEFAULT_OLLAMA_BASE_URL,
}
.to_string()
});
@@ -1229,6 +1264,7 @@ impl Config {
ApiProvider::Fireworks => "fireworks",
ApiProvider::Sglang => "sglang",
ApiProvider::Vllm => "vllm",
ApiProvider::Ollama => "ollama",
};
// 0. Explicit in-memory override (set by onboarding / provider
@@ -1289,10 +1325,9 @@ impl Config {
"Fireworks AI API key not found. Run 'deepseek auth set --provider fireworks', \
set FIREWORKS_API_KEY, or add [providers.fireworks] api_key in ~/.deepseek/config.toml."
),
// Self-hosted SGLang deployments commonly run without auth on
// localhost. Return an empty key and let the client omit the
// Authorization header.
ApiProvider::Sglang | ApiProvider::Vllm => Ok(String::new()),
// Self-hosted deployments commonly run without auth on localhost.
// Return an empty key and let the client omit the Authorization header.
ApiProvider::Sglang | ApiProvider::Vllm | ApiProvider::Ollama => Ok(String::new()),
}
}
@@ -1842,11 +1877,22 @@ fn apply_env_overrides(config: &mut Config) {
ApiProvider::Fireworks => &mut providers.fireworks,
ApiProvider::Sglang => &mut providers.sglang,
ApiProvider::Vllm => &mut providers.vllm,
ApiProvider::Ollama => &mut providers.ollama,
};
let mut provider_headers = entry.http_headers.clone().unwrap_or_default();
provider_headers.extend(headers);
entry.http_headers = Some(provider_headers);
}
if matches!(config.api_provider(), ApiProvider::Ollama)
&& let Ok(value) = std::env::var("OLLAMA_BASE_URL")
&& !value.trim().is_empty()
{
config
.providers
.get_or_insert_with(ProvidersConfig::default)
.ollama
.base_url = Some(value);
}
if matches!(config.api_provider(), ApiProvider::Sglang)
&& let Ok(value) = std::env::var("SGLANG_MODEL")
{
@@ -1857,6 +1903,11 @@ fn apply_env_overrides(config: &mut Config) {
{
config.default_text_model = Some(value);
}
if matches!(config.api_provider(), ApiProvider::Ollama)
&& let Ok(value) = std::env::var("OLLAMA_MODEL")
{
config.default_text_model = Some(value);
}
if let Ok(value) =
std::env::var("DEEPSEEK_MODEL").or_else(|_| std::env::var("DEEPSEEK_DEFAULT_TEXT_MODEL"))
{
@@ -2035,6 +2086,7 @@ fn apply_env_overrides(config: &mut Config) {
fn normalize_model_config(config: &mut Config) {
if let Some(model) = config.default_text_model.as_deref()
&& !matches!(config.api_provider(), ApiProvider::Ollama)
&& let Some(normalized) = normalize_model_for_provider(config.api_provider(), model)
{
config.default_text_model = Some(normalized);
@@ -2085,6 +2137,9 @@ fn normalize_model_config(config: &mut Config) {
}
fn normalize_model_for_provider(provider: ApiProvider, model: &str) -> Option<String> {
if matches!(provider, ApiProvider::Ollama) {
return None;
}
normalize_model_name(model).map(|normalized| model_for_provider(provider, normalized))
}
@@ -2274,6 +2329,7 @@ fn merge_providers(
fireworks: merge_provider_config(base.fireworks, override_cfg.fireworks),
sglang: merge_provider_config(base.sglang, override_cfg.sglang),
vllm: merge_provider_config(base.vllm, override_cfg.vllm),
ollama: merge_provider_config(base.ollama, override_cfg.ollama),
}),
}
}
@@ -2692,6 +2748,7 @@ pub fn active_provider_has_env_api_key(config: &Config) -> bool {
}
ApiProvider::Sglang => std::env::var("SGLANG_API_KEY").is_ok_and(|k| !k.trim().is_empty()),
ApiProvider::Vllm => std::env::var("VLLM_API_KEY").is_ok_and(|k| !k.trim().is_empty()),
ApiProvider::Ollama => std::env::var("OLLAMA_API_KEY").is_ok_and(|k| !k.trim().is_empty()),
}
}
@@ -2713,6 +2770,7 @@ pub fn has_api_key_for(config: &Config, provider: ApiProvider) -> bool {
ApiProvider::Fireworks => "FIREWORKS_API_KEY",
ApiProvider::Sglang => "SGLANG_API_KEY",
ApiProvider::Vllm => "VLLM_API_KEY",
ApiProvider::Ollama => "OLLAMA_API_KEY",
};
if std::env::var(env_var).is_ok_and(|k| !k.trim().is_empty()) {
return true;
@@ -2723,8 +2781,11 @@ pub fn has_api_key_for(config: &Config, provider: ApiProvider) -> bool {
return true;
}
// SGLang is self-hosted and typically runs without authentication.
if matches!(provider, ApiProvider::Sglang | ApiProvider::Vllm) {
// Self-hosted providers typically run without authentication.
if matches!(
provider,
ApiProvider::Sglang | ApiProvider::Vllm | ApiProvider::Ollama
) {
return true;
}
@@ -2776,6 +2837,7 @@ pub fn save_api_key_for(provider: ApiProvider, api_key: &str) -> Result<PathBuf>
ApiProvider::Fireworks => "providers.fireworks",
ApiProvider::Sglang => "providers.sglang",
ApiProvider::Vllm => "providers.vllm",
ApiProvider::Ollama => "providers.ollama",
};
// Parse existing TOML (or start fresh) so we can edit the right table
@@ -2808,6 +2870,7 @@ pub fn save_api_key_for(provider: ApiProvider, api_key: &str) -> Result<PathBuf>
ApiProvider::Fireworks => "fireworks",
ApiProvider::Sglang => "sglang",
ApiProvider::Vllm => "vllm",
ApiProvider::Ollama => "ollama",
};
let entry = providers
.entry(key_inside.to_string())
@@ -2927,6 +2990,9 @@ mod tests {
vllm_api_key: Option<OsString>,
vllm_base_url: Option<OsString>,
vllm_model: Option<OsString>,
ollama_api_key: Option<OsString>,
ollama_base_url: Option<OsString>,
ollama_model: Option<OsString>,
}
impl EnvGuard {
@@ -2961,6 +3027,9 @@ mod tests {
let vllm_api_key_prev = env::var_os("VLLM_API_KEY");
let vllm_base_url_prev = env::var_os("VLLM_BASE_URL");
let vllm_model_prev = env::var_os("VLLM_MODEL");
let ollama_api_key_prev = env::var_os("OLLAMA_API_KEY");
let ollama_base_url_prev = env::var_os("OLLAMA_BASE_URL");
let ollama_model_prev = env::var_os("OLLAMA_MODEL");
// Safety: test-only environment mutation guarded by a global mutex.
unsafe {
env::set_var("HOME", &home_str);
@@ -2990,6 +3059,9 @@ mod tests {
env::remove_var("VLLM_API_KEY");
env::remove_var("VLLM_BASE_URL");
env::remove_var("VLLM_MODEL");
env::remove_var("OLLAMA_API_KEY");
env::remove_var("OLLAMA_BASE_URL");
env::remove_var("OLLAMA_MODEL");
}
Self {
home: home_prev,
@@ -3019,6 +3091,9 @@ mod tests {
vllm_api_key: vllm_api_key_prev,
vllm_base_url: vllm_base_url_prev,
vllm_model: vllm_model_prev,
ollama_api_key: ollama_api_key_prev,
ollama_base_url: ollama_base_url_prev,
ollama_model: ollama_model_prev,
}
}
}
@@ -3057,6 +3132,9 @@ mod tests {
Self::restore_var("VLLM_API_KEY", self.vllm_api_key.take());
Self::restore_var("VLLM_BASE_URL", self.vllm_base_url.take());
Self::restore_var("VLLM_MODEL", self.vllm_model.take());
Self::restore_var("OLLAMA_API_KEY", self.ollama_api_key.take());
Self::restore_var("OLLAMA_BASE_URL", self.ollama_base_url.take());
Self::restore_var("OLLAMA_MODEL", self.ollama_model.take());
}
}
}
@@ -4221,6 +4299,97 @@ http_headers = { "X-Model-Provider-Id" = "from-file" }
Ok(())
}
#[test]
fn ollama_provider_uses_local_defaults_without_api_key() -> Result<()> {
let _lock = lock_test_env();
let nanos = SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap()
.as_nanos();
let temp_root = env::temp_dir().join(format!(
"deepseek-tui-ollama-defaults-{}-{}",
std::process::id(),
nanos
));
fs::create_dir_all(&temp_root)?;
let _guard = EnvGuard::new(&temp_root);
let config = Config {
provider: Some("ollama".to_string()),
..Default::default()
};
config.validate()?;
assert_eq!(config.api_provider(), ApiProvider::Ollama);
assert_eq!(config.default_model(), DEFAULT_OLLAMA_MODEL);
assert_eq!(config.deepseek_base_url(), DEFAULT_OLLAMA_BASE_URL);
assert_eq!(config.deepseek_api_key()?, "");
assert!(has_api_key_for(&config, ApiProvider::Ollama));
Ok(())
}
#[test]
fn ollama_model_is_passed_through_verbatim() -> Result<()> {
let _lock = lock_test_env();
let nanos = SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap()
.as_nanos();
let temp_root = env::temp_dir().join(format!(
"deepseek-tui-ollama-model-test-{}-{}",
std::process::id(),
nanos
));
fs::create_dir_all(&temp_root)?;
let _guard = EnvGuard::new(&temp_root);
let config_path = temp_root.join(".deepseek").join("config.toml");
ensure_parent_dir(&config_path)?;
fs::write(
&config_path,
r#"provider = "ollama"
[providers.ollama]
base_url = "http://127.0.0.1:11434/v1"
model = "qwen2.5-coder:7b"
"#,
)?;
let config = Config::load(None, None)?;
assert_eq!(config.api_provider(), ApiProvider::Ollama);
assert_eq!(config.default_model(), "qwen2.5-coder:7b");
assert_eq!(config.deepseek_base_url(), "http://127.0.0.1:11434/v1");
Ok(())
}
#[test]
fn ollama_env_overrides_base_url_and_model() -> Result<()> {
let _lock = lock_test_env();
let nanos = SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap()
.as_nanos();
let temp_root = env::temp_dir().join(format!(
"deepseek-tui-ollama-env-test-{}-{}",
std::process::id(),
nanos
));
fs::create_dir_all(&temp_root)?;
let _guard = EnvGuard::new(&temp_root);
// Safety: test-only environment mutation guarded by a global mutex.
unsafe {
env::set_var("DEEPSEEK_PROVIDER", "ollama-local");
env::set_var("OLLAMA_BASE_URL", "http://ollama.example/v1");
env::set_var("OLLAMA_MODEL", "deepseek-coder-v2:16b");
}
let config = Config::load(None, None)?;
assert_eq!(config.api_provider(), ApiProvider::Ollama);
assert_eq!(config.deepseek_base_url(), "http://ollama.example/v1");
assert_eq!(config.default_model(), "deepseek-coder-v2:16b");
Ok(())
}
#[test]
fn openrouter_env_api_key_resolves_via_deepseek_api_key() -> Result<()> {
let _lock = lock_test_env();
@@ -4685,6 +4854,19 @@ model = "deepseek-v4-pro"
assert!(!cap.cache_telemetry_supported);
}
#[test]
fn provider_capability_ollama_is_openai_compatible_without_thinking() {
let cap = provider_capability(ApiProvider::Ollama, "deepseek-v3.1:671b");
assert_eq!(cap.context_window, 8192);
assert_eq!(cap.max_output, 4096);
assert!(!cap.thinking_supported);
assert!(!cap.cache_telemetry_supported);
assert_eq!(
cap.request_payload_mode,
RequestPayloadMode::ChatCompletions
);
}
#[test]
fn provider_capability_non_v4_model_has_smaller_window() {
let cap = provider_capability(ApiProvider::Deepseek, "deepseek-coder");
crates/tui/src/core/engine.rs
+1
View File
@@ -375,6 +375,7 @@ impl Engine {
ApiProvider::Fireworks => "FIREWORKS_API_KEY",
ApiProvider::Sglang => "SGLANG_API_KEY",
ApiProvider::Vllm => "VLLM_API_KEY",
ApiProvider::Ollama => "OLLAMA_API_KEY",
};
Some(format!(
crates/tui/src/core/engine/tests.rs
+5 -6
View File
@@ -1,6 +1,7 @@
use super::*;
use crate::models::SystemBlock;
use crate::test_support::lock_test_env;
use serde_json::json;
use std::collections::HashSet;
use std::ffi::OsString;
@@ -13,8 +14,6 @@ use tempfile::tempdir;
const WORKING_SET_SUMMARY_MARKER: &str = "## Repo Working Set";
static CAPACITY_MEMORY_ENV_LOCK: LazyLock<tokio::sync::Mutex<()>> =
LazyLock::new(|| tokio::sync::Mutex::new(()));
static API_KEY_ENV_LOCK: LazyLock<std::sync::Mutex<()>> =
LazyLock::new(|| std::sync::Mutex::new(()));
struct ScopedCapacityMemoryDir {
previous: Option<OsString>,
@@ -52,7 +51,7 @@ struct ScopedDeepSeekApiKey {
impl ScopedDeepSeekApiKey {
fn set(value: &str) -> Self {
let previous = std::env::var_os("DEEPSEEK_API_KEY");
// Safety: tests using this helper serialize with API_KEY_ENV_LOCK and
// Safety: tests using this helper serialize with lock_test_env() and
// restore the original value in Drop.
unsafe {
std::env::set_var("DEEPSEEK_API_KEY", value);
@@ -63,7 +62,7 @@ impl ScopedDeepSeekApiKey {
impl Drop for ScopedDeepSeekApiKey {
fn drop(&mut self) {
// Safety: tests using this helper serialize with API_KEY_ENV_LOCK.
// Safety: tests using this helper serialize with lock_test_env().
unsafe {
if let Some(previous) = self.previous.take() {
std::env::set_var("DEEPSEEK_API_KEY", previous);
@@ -85,7 +84,7 @@ fn build_engine_with_capacity(capacity: CapacityControllerConfig) -> Engine {
#[test]
fn env_only_auth_error_gets_recovery_hint() {
let _guard = API_KEY_ENV_LOCK.lock().unwrap_or_else(|e| e.into_inner());
let _guard = lock_test_env();
let _env = ScopedDeepSeekApiKey::set("stale-env-key");
let (engine, _handle) = Engine::new(EngineConfig::default(), &Config::default());
@@ -99,7 +98,7 @@ fn env_only_auth_error_gets_recovery_hint() {
#[test]
fn config_auth_error_does_not_blame_env() {
let _guard = API_KEY_ENV_LOCK.lock().unwrap_or_else(|e| e.into_inner());
let _guard = lock_test_env();
let _env = ScopedDeepSeekApiKey::set("stale-env-key");
let cfg = Config {
api_key: Some("fresh-config-key".to_string()),
crates/tui/src/localization.rs
+6 -4
View File
@@ -756,7 +756,7 @@ fn english(id: MessageId) -> &'static str {
"Switch to plan mode and review suggested implementation steps"
}
MessageId::CmdProviderDescription => {
"Switch or view the active LLM backend (deepseek | nvidia-nim)"
"Switch or view the active LLM backend (deepseek | nvidia-nim | ollama)"
}
MessageId::CmdQueueDescription => "View or edit queued messages",
MessageId::CmdRecallDescription => "Search prior cycle archives (BM25 over message text)",
@@ -1036,7 +1036,7 @@ fn japanese(id: MessageId) -> Option<&'static str> {
MessageId::CmdNoteDescription => "永続ノートファイル（.deepseek/notes.md）に追記",
MessageId::CmdPlanDescription => "Plan モードに切り替え、推奨される実装手順を確認",
MessageId::CmdProviderDescription => {
"現在の LLM バックエンドを切り替え・確認（deepseek | nvidia-nim）"
"現在の LLM バックエンドを切り替え・確認（deepseek | nvidia-nim | ollama）"
}
MessageId::CmdQueueDescription => "キューされたメッセージを確認・編集",
MessageId::CmdRecallDescription => {
@@ -1299,7 +1299,9 @@ fn chinese_simplified(id: MessageId) -> Option<&'static str> {
MessageId::CmdNetworkDescription => "管理网络允许和拒绝规则",
MessageId::CmdNoteDescription => "将笔记追加到持久笔记文件（.deepseek/notes.md）",
MessageId::CmdPlanDescription => "切换到 Plan 模式并查看建议的实现步骤",
MessageId::CmdProviderDescription => "切换或查看当前 LLM 后端（deepseek | nvidia-nim）",
MessageId::CmdProviderDescription => {
"切换或查看当前 LLM 后端（deepseek | nvidia-nim | ollama）"
}
MessageId::CmdQueueDescription => "查看或编辑已排队的消息",
MessageId::CmdRecallDescription => "搜索此前的循环归档（基于消息文本的 BM25 检索）",
MessageId::CmdRenameDescription => "重命名当前会话",
@@ -1558,7 +1560,7 @@ fn portuguese_brazil(id: MessageId) -> Option<&'static str> {
"Mudar para o modo plan e revisar os passos de implementação sugeridos"
}
MessageId::CmdProviderDescription => {
"Trocar ou exibir o backend LLM ativo (deepseek | nvidia-nim)"
"Trocar ou exibir o backend LLM ativo (deepseek | nvidia-nim | ollama)"
}
MessageId::CmdQueueDescription => "Ver ou editar mensagens enfileiradas",
MessageId::CmdRecallDescription => {
crates/tui/src/main.rs
+19
View File
@@ -1355,6 +1355,9 @@ fn run_setup_status(config: &Config, workspace: &Path) -> Result<()> {
"VLLM_API_KEY",
"deepseek auth set --provider vllm --api-key \"...\"",
),
crate::config::ApiProvider::Ollama => {
("OLLAMA_API_KEY", "deepseek auth set --provider ollama")
}
crate::config::ApiProvider::Deepseek | crate::config::ApiProvider::DeepseekCN => {
("DEEPSEEK_API_KEY", "deepseek auth set --provider deepseek")
}
@@ -1369,6 +1372,7 @@ fn run_setup_status(config: &Config, workspace: &Path) -> Result<()> {
crate::config::ApiProvider::Fireworks => "fireworks",
crate::config::ApiProvider::Sglang => "sglang",
crate::config::ApiProvider::Vllm => "vllm",
crate::config::ApiProvider::Ollama => "ollama",
crate::config::ApiProvider::Deepseek
| crate::config::ApiProvider::DeepseekCN => "deepseek",
}
@@ -1605,6 +1609,11 @@ async fn run_doctor(config: &Config, workspace: &Path, config_path_override: Opt
"vllm",
&["VLLM_API_KEY"][..],
),
(
crate::config::ApiProvider::Ollama,
"ollama",
&["OLLAMA_API_KEY"][..],
),
] {
let in_env = env_names.iter().any(|n| {
std::env::var(n)
@@ -1646,6 +1655,16 @@ async fn run_doctor(config: &Config, workspace: &Path, config_path_override: Opt
ApiKeySource::Config => "config.toml",
ApiKeySource::Keyring => "OS keyring",
ApiKeySource::Env => "environment",
ApiKeySource::Missing
if matches!(
config.api_provider(),
crate::config::ApiProvider::Sglang
| crate::config::ApiProvider::Vllm
| crate::config::ApiProvider::Ollama
) =>
{
"optional local auth"
}
ApiKeySource::Missing => "unknown source",
};
println!(
crates/tui/src/tui/provider_picker.rs
+36 -8
View File
@@ -1,5 +1,5 @@
//! `/provider` picker modal — pick a provider (DeepSeek / NVIDIA NIM /
//! OpenRouter / Novita) and, if it lacks credentials, type the API key
//! hosted providers / self-hosted providers) and, if it lacks credentials, type the API key
//! inline before completing the switch (#52).
//!
//! The picker is intentionally a single modal with two visible states:
@@ -93,6 +93,19 @@ impl ProviderPickerView {
ApiProvider::Fireworks => "FIREWORKS_API_KEY",
ApiProvider::Sglang => "SGLANG_API_KEY",
ApiProvider::Vllm => "VLLM_API_KEY",
ApiProvider::Ollama => "OLLAMA_API_KEY",
}
}
fn provider_hint(provider: ApiProvider, has_key: bool) -> String {
match provider {
ApiProvider::Ollama => "self-hosted; defaults to http://localhost:11434".to_string(),
ApiProvider::Sglang | ApiProvider::Vllm if has_key => {
"(configured; optional key)".to_string()
}
ApiProvider::Sglang | ApiProvider::Vllm => "(optional key)".to_string(),
_ if has_key => "(configured)".to_string(),
_ => "(needs API key)".to_string(),
}
}
@@ -141,11 +154,7 @@ impl ProviderPickerView {
} else {
Style::default().fg(palette::STATUS_WARNING)
};
let hint = if *has_key {
"(configured)".to_string()
} else {
"(needs API key)".to_string()
};
let hint = Self::provider_hint(*provider, *has_key);
lines.push(Line::from(vec![
Span::raw(" "),
Span::styled(arrow, label_style),
@@ -358,7 +367,7 @@ mod tests {
}
#[test]
fn picker_lists_all_seven_providers() {
fn picker_lists_all_providers() {
let config = Config::default();
let picker = ProviderPickerView::new(ApiProvider::Deepseek, &config);
let names: Vec<_> = picker
@@ -376,11 +385,30 @@ mod tests {
"Novita AI",
"Fireworks AI",
"SGLang",
"vLLM"
"vLLM",
"Ollama"
]
);
}
#[test]
fn ollama_is_selectable_without_key() {
let config = Config::default();
let mut picker = ProviderPickerView::new(ApiProvider::Deepseek, &config);
for _ in 0..8 {
picker.handle_key(key(KeyCode::Down));
}
assert_eq!(picker.selected_provider(), ApiProvider::Ollama);
assert!(picker.selected_has_key());
let action = picker.handle_key(key(KeyCode::Enter));
match action {
ViewAction::EmitAndClose(ViewEvent::ProviderPickerApplied { provider }) => {
assert_eq!(provider, ApiProvider::Ollama);
}
other => panic!("expected ProviderPickerApplied, got {other:?}"),
}
}
#[test]
fn picker_marks_active_provider_as_initial_selection() {
let config = Config::default();
crates/tui/src/tui/ui.rs
+3
View File
@@ -4698,6 +4698,7 @@ async fn execute_command_input(
providers.fireworks.api_key = None;
providers.sglang.api_key = None;
providers.vllm.api_key = None;
providers.ollama.api_key = None;
}
app.api_key_env_only = crate::config::active_provider_uses_env_only_api_key(config);
}
@@ -5076,6 +5077,7 @@ fn render(f: &mut Frame, app: &mut App) {
crate::config::ApiProvider::Fireworks => Some("Fireworks"),
crate::config::ApiProvider::Sglang => Some("SGLang"),
crate::config::ApiProvider::Vllm => Some("vLLM"),
crate::config::ApiProvider::Ollama => Some("Ollama"),
};
let header_data = HeaderData::new(
app.mode,
@@ -5706,6 +5708,7 @@ async fn apply_provider_picker_api_key(
ApiProvider::Fireworks => &mut providers.fireworks,
ApiProvider::Sglang => &mut providers.sglang,
ApiProvider::Vllm => &mut providers.vllm,
ApiProvider::Ollama => &mut providers.ollama,
};
entry.api_key = Some(api_key);
}
docs/CONFIGURATION.md
+17 -8
View File
@@ -55,14 +55,15 @@ the legacy `deepseek login --api-key ...` alias) saves the key to
`~/.deepseek/config.toml`, and `deepseek --model deepseek-v4-flash` is forwarded
to the TUI as `DEEPSEEK_MODEL`.
For hosted or self-hosted DeepSeek V4 providers, set `provider = "nvidia-nim"`,
`"fireworks"`, `"sglang"`, or `"vllm"` or pass `deepseek --provider <name>`. The facade
For hosted or self-hosted providers, set `provider = "nvidia-nim"`,
`"fireworks"`, `"sglang"`, `"vllm"`, or `"ollama"` or pass `deepseek --provider <name>`. The facade
saves provider credentials to the shared user config and forwards the resolved
key, base URL, provider, and model to the TUI process. Use
`deepseek auth set --provider nvidia-nim --api-key "YOUR_NVIDIA_API_KEY"` or
`deepseek auth set --provider fireworks --api-key "YOUR_FIREWORKS_API_KEY"` to
save hosted-provider keys through the facade. SGLang and vLLM are self-hosted and can run
without an API key by default.
save hosted-provider keys through the facade. SGLang, vLLM, and Ollama are self-hosted and can run
without an API key by default. Ollama defaults to `http://localhost:11434/v1` and sends model tags
such as `deepseek-coder:1.3b` or `qwen2.5-coder:7b` unchanged.
Third-party OpenAI-compatible gateways that need extra request headers can set
`http_headers = { "X-Model-Provider-Id" = "your-model-provider" }` at the top
@@ -114,6 +115,11 @@ default_text_model = "deepseek-ai/DeepSeek-V4-Pro"
provider = "vllm"
base_url = "http://localhost:8000/v1"
default_text_model = "deepseek-ai/DeepSeek-V4-Pro"
[profiles.ollama]
provider = "ollama"
base_url = "http://localhost:11434/v1"
default_text_model = "deepseek-coder:1.3b"
```
Select a profile with:
@@ -130,7 +136,7 @@ These override config values:
- `DEEPSEEK_API_KEY`
- `DEEPSEEK_BASE_URL`
- `DEEPSEEK_HTTP_HEADERS` (custom model request headers, comma-separated `name=value` pairs)
- `DEEPSEEK_PROVIDER` (`deepseek|nvidia-nim|openrouter|novita|fireworks|sglang|vllm`)
- `DEEPSEEK_PROVIDER` (`deepseek|nvidia-nim|openrouter|novita|fireworks|sglang|vllm|ollama`)
- `DEEPSEEK_MODEL` or `DEEPSEEK_DEFAULT_TEXT_MODEL`
- `NVIDIA_API_KEY` or `NVIDIA_NIM_API_KEY` (preferred when provider is `nvidia-nim`; falls back to `DEEPSEEK_API_KEY`)
- `NVIDIA_NIM_BASE_URL`, `NIM_BASE_URL`, or `NVIDIA_BASE_URL`
@@ -143,6 +149,9 @@ These override config values:
- `VLLM_BASE_URL`
- `VLLM_MODEL`
- `VLLM_API_KEY` (optional; many localhost vLLM servers do not require auth)
- `OLLAMA_BASE_URL`
- `OLLAMA_MODEL`
- `OLLAMA_API_KEY` (optional; many localhost Ollama servers do not require auth)
- `DEEPSEEK_LOG_LEVEL` or `RUST_LOG` (`info`/`debug`/`trace` enables lightweight verbose logs)
- `DEEPSEEK_SKILLS_DIR`
- `DEEPSEEK_MCP_CONFIG`
@@ -316,10 +325,10 @@ If you are upgrading from older releases:
### Core keys (used by the TUI/engine)
- `provider` (string, optional): `deepseek` (default), `deepseek-cn`, `nvidia-nim`, `openrouter`, `novita`, `fireworks`, `sglang`, or `vllm`. `deepseek-cn` uses DeepSeek's mainland China endpoint (`https://api.deepseeki.com`); `nvidia-nim` targets NVIDIA's NIM-hosted DeepSeek endpoints through `https://integrate.api.nvidia.com/v1`; `fireworks` targets `https://api.fireworks.ai/inference/v1`; `sglang` targets a self-hosted OpenAI-compatible endpoint, defaulting to `http://localhost:30000/v1`; `vllm` targets a self-hosted vLLM OpenAI-compatible endpoint, defaulting to `http://localhost:8000/v1`.
- `api_key` (string, required): must be non-empty (or set `DEEPSEEK_API_KEY`).
- `provider` (string, optional): `deepseek` (default), `deepseek-cn`, `nvidia-nim`, `openrouter`, `novita`, `fireworks`, `sglang`, `vllm`, or `ollama`. `deepseek-cn` uses DeepSeek's mainland China endpoint (`https://api.deepseeki.com`); `nvidia-nim` targets NVIDIA's NIM-hosted DeepSeek endpoints through `https://integrate.api.nvidia.com/v1`; `fireworks` targets `https://api.fireworks.ai/inference/v1`; `sglang` targets a self-hosted OpenAI-compatible endpoint, defaulting to `http://localhost:30000/v1`; `vllm` targets a self-hosted vLLM OpenAI-compatible endpoint, defaulting to `http://localhost:8000/v1`; `ollama` targets Ollama's OpenAI-compatible endpoint, defaulting to `http://localhost:11434/v1`.
- `api_key` (string, required for hosted providers): must be non-empty for DeepSeek/hosted providers (or set the provider API key env var). Self-hosted SGLang, vLLM, and Ollama can omit it.
- `base_url` (string, optional): defaults to `https://api.deepseek.com` for DeepSeek's OpenAI-compatible Chat Completions API, `https://api.deepseeki.com` for `provider = "deepseek-cn"`, or the provider-specific endpoint for hosted/self-hosted providers. `https://api.deepseek.com/v1` is also accepted for SDK compatibility; use `https://api.deepseek.com/beta` only for DeepSeek beta features such as strict tool mode, chat prefix completion, and FIM completion.
- `default_text_model` (string, optional): defaults to `deepseek-v4-pro` for DeepSeek, `deepseek-ai/deepseek-v4-pro` for NVIDIA NIM, `accounts/fireworks/models/deepseek-v4-pro` for Fireworks, and `deepseek-ai/DeepSeek-V4-Pro` for SGLang. Current public DeepSeek IDs are `deepseek-v4-pro` and `deepseek-v4-flash`, both with 1M context windows and thinking mode enabled by default. Legacy `deepseek-chat` and `deepseek-reasoner` remain compatibility aliases for `deepseek-v4-flash`. Provider-specific mappings translate `deepseek-v4-pro` / `deepseek-v4-flash` to each provider's model ID where supported. Use `/models` or `deepseek models` to discover live IDs from your configured endpoint. `DEEPSEEK_MODEL` overrides this for a single process.
- `default_text_model` (string, optional): defaults to `deepseek-v4-pro` for DeepSeek, `deepseek-ai/deepseek-v4-pro` for NVIDIA NIM, `accounts/fireworks/models/deepseek-v4-pro` for Fireworks, `deepseek-ai/DeepSeek-V4-Pro` for SGLang/vLLM, and `deepseek-coder:1.3b` for Ollama. Current public DeepSeek IDs are `deepseek-v4-pro` and `deepseek-v4-flash`, both with 1M context windows and thinking mode enabled by default. Legacy `deepseek-chat` and `deepseek-reasoner` remain compatibility aliases for `deepseek-v4-flash`. Provider-specific mappings translate `deepseek-v4-pro` / `deepseek-v4-flash` to each provider's model ID where supported. Ollama model tags are passed through unchanged. Use `/models` or `deepseek models` to discover live IDs from your configured endpoint. `DEEPSEEK_MODEL` overrides this for a single process.
- `reasoning_effort` (string, optional): `off`, `low`, `medium`, `high`, or `max`; defaults to the configured UI tier. DeepSeek Platform receives top-level `thinking` / `reasoning_effort` fields. NVIDIA NIM receives equivalent settings through `chat_template_kwargs`.
- `allow_shell` (bool, optional): defaults to `true` (sandboxed).
- `approval_policy` (string, optional): `on-request`, `untrusted`, or `never`. Runtime `approval_mode` editing in `/config` also accepts `on-request` and `untrusted` aliases.