Solution for long term memory for agent coding CLIs and to facilitate handoff between different agent vendors
Rust
fix(llm): apply llm_base_url to anthropic and openai providers
lhzapata1ff8c618d736
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex 59954396..a24de7fd 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -11,6 +11,16 @@ version in [`docs/changelog/`](docs/changelog/):## [Unreleased]+### Fixed+- Fixed an explicit `llm_base_url` (or `AI_MEMORY_LLM_BASE_URL`) being dropped+ for the `anthropic` and `openai` providers. Config resolved the override for+ every provider, as documented, but the provider factory never applied it to+ those two, so requests and the API key went straight to api.anthropic.com or+ api.openai.com with nothing in the log. A gateway or reverse proxy set in front+ of either vendor now receives the traffic. The docs now name the backends that+ take no override: the ChatGPT backend behind `openai-oauth` and `codex`, and+ `cursor`'s CLI. (#1347)+## [3.1.0] - 2026-10-16### Addeddiff --git a/crates/ai-memory-cli/src/config.rs b/crates/ai-memory-cli/src/config.rsindex 39ac68c8..d924a019 100644--- a/crates/ai-memory-cli/src/config.rs+++ b/crates/ai-memory-cli/src/config.rs@@ -3134,9 +3134,10 @@ impl Config {/// Base URL for `provider`, from the two sources that can supply one.////// An explicit ai-memory setting (`llm_base_url` in `config.toml`, or- /// `AI_MEMORY_LLM_BASE_URL`) configures any provider: naming ai-memory is- /// how an operator says they mean it, and proxying a vendor endpoint is a- /// legitimate deployment.+ /// `AI_MEMORY_LLM_BASE_URL`) configures every HTTP API provider: naming+ /// ai-memory is how an operator says they mean it, and proxying a vendor+ /// endpoint is a legitimate deployment. The ChatGPT backend+ /// (`openai-oauth`, `codex`) and `cursor`'s CLI take no override.////// The bare `LLM_BASE_URL` is different in kind — a cross-tool convention/// an operator exports once for a local Ollama and forgets. It reachesdiff --git a/crates/ai-memory-llm/src/factory.rs b/crates/ai-memory-llm/src/factory.rsindex 451f4527..91c59fc8 100644--- a/crates/ai-memory-llm/src/factory.rs+++ b/crates/ai-memory-llm/src/factory.rs@@ -384,8 +384,12 @@ pub fn build_provider(config: ProviderConfig) -> LlmResult<Arc<dyn LlmProvider>>match config.provider {ProviderChoice::Anthropic => {let key = config.auth.require_api_key()?;+ let mut provider = AnthropicProvider::new(key, config.model)?;+ if let Some(url) = config.base_url {+ provider = provider.with_base_url(url);+ }Ok(Arc::new(- AnthropicProvider::new(key, config.model)?+ provider.with_timeout_secs(timeout).with_reasoning_effort(config.reasoning_effort).with_extra_headers(extra_headers),@@ -393,8 +397,12 @@ pub fn build_provider(config: ProviderConfig) -> LlmResult<Arc<dyn LlmProvider>>}ProviderChoice::OpenAi => {let key = config.auth.require_api_key()?;+ let mut provider = OpenAiProvider::new(key, config.model)?;+ if let Some(url) = config.base_url {+ provider = provider.with_base_url(url);+ }Ok(Arc::new(- OpenAiProvider::new(key, config.model)?+ provider.with_timeout_secs(timeout).with_reasoning_effort(config.reasoning_effort).with_extra_headers(extra_headers),diff --git a/crates/ai-memory-llm/tests/suite/extra_headers_on_the_wire.rs b/crates/ai-memory-llm/tests/suite/extra_headers_on_the_wire.rsindex b78c700b..70bc57cb 100644--- a/crates/ai-memory-llm/tests/suite/extra_headers_on_the_wire.rs+++ b/crates/ai-memory-llm/tests/suite/extra_headers_on_the_wire.rs@@ -231,6 +231,89 @@ async fn an_operator_base_url_redirects_the_opencode_provider() {assert_eq!(values(request, "user-agent"), vec![DEFAULT_USER_AGENT]);}+/// An explicit `llm_base_url` configures the vendor-fixed providers too: it is+/// how an operator routes `anthropic` or `openai` through a gateway on+/// purpose. Dropping it sends the request, and the key, straight to the vendor+/// with nothing in the log.+#[tokio::test]+async fn an_operator_base_url_redirects_the_anthropic_provider() {+ let server = MockServer::start().await;+ Mock::given(method("POST"))+ .and(path("/v1/messages"))+ .respond_with(ResponseTemplate::new(200).set_body_json(json!({+ "model": "claude-haiku-4-5",+ "content": [{"type": "text", "text": "ok"}],+ "usage": {"input_tokens": 1, "output_tokens": 1},+ })))+ .mount(&server)+ .await;++ let provider = build_provider(ProviderConfig {+ provider: ProviderChoice::Anthropic,+ model: "claude-haiku-4-5".into(),+ auth: ProviderAuth::required_api_key_from_env(+ "ANTHROPIC_API_KEY",+ Some(SecretString::from("sk-ant-test")),+ ),+ base_url: Some(server.uri()),+ compat_strict: false,+ compat_disable_thinking: false,+ request_timeout_secs: 30,+ reasoning_effort: None,+ extra_headers: ExtraHeaders::default(),+ })+ .expect("provider builds");++ provider+ .complete(ChatRequest::user_prompt("hi"))+ .await+ .expect("the override must reach the mock, not api.anthropic.com");++ let received = server+ .received_requests()+ .await+ .expect("mock recorded requests");+ assert_eq!(received.len(), 1, "request did not reach the override");+}++/// Same contract for `openai`, whose default is api.openai.com.+#[tokio::test]+async fn an_operator_base_url_redirects_the_openai_provider() {+ let server = MockServer::start().await;+ Mock::given(method("POST"))+ .and(path("/v1/chat/completions"))+ .respond_with(ResponseTemplate::new(200).set_body_json(ok_body()))+ .mount(&server)+ .await;++ let provider = build_provider(ProviderConfig {+ provider: ProviderChoice::OpenAi,+ model: "gpt-5.4-mini".into(),+ auth: ProviderAuth::required_api_key_from_env(+ "OPENAI_API_KEY",+ Some(SecretString::from("sk-test")),+ ),+ base_url: Some(server.uri()),+ compat_strict: false,+ compat_disable_thinking: false,+ request_timeout_secs: 30,+ reasoning_effort: None,+ extra_headers: ExtraHeaders::default(),+ })+ .expect("provider builds");++ provider+ .complete(ChatRequest::user_prompt("hi"))+ .await+ .expect("the override must reach the mock, not api.openai.com");++ let received = server+ .received_requests()+ .await+ .expect("mock recorded requests");+ assert_eq!(received.len(), 1, "request did not reach the override");+}+/// The load-bearing claim of layering operator headers over the provider's/// own defaults: an explicit `AI_MEMORY_LLM_HEADERS` entry must beat the/// per-operation id, and must arrive exactly once. Appending instead ofdiff --git a/docs/ARCHITECTURE.md b/docs/ARCHITECTURE.mdindex b5291550..ce0816a3 100644--- a/docs/ARCHITECTURE.md+++ b/docs/ARCHITECTURE.md@@ -1307,8 +1307,10 @@ ANTHROPIC_API_KEY / OPENAI_API_KEY / GEMINI_API_KEY / LLM_API_KEYAI_MEMORY_LLM_BASE_URL required for openai-compat (Ollama, vLLM); optional override foropencode (defaults to the Go endpoint, sethttps://opencode.ai/zen/v1 for Zen's catalogue).- Applies to any provider: naming ai-memory is how an- operator says a vendor endpoint is proxied on purpose+ Applies to every HTTP API provider: naming ai-memory+ is how an operator says a vendor endpoint is proxied+ on purpose. The ChatGPT backend (openai-oauth, codex)+ and cursor's CLI have no overrideLLM_BASE_URL the unprefixed cross-tool convention, accepted foropenai-compat and opencode only. Providers with afixed vendor endpoint (anthropic, openai, gemini,