refactor(providers): separate out api implementation in separate crate

Cargo.toml

@@ -1,5 +1,5 @@
 [workspace]
-members = ["crates/rullm-core", "crates/rullm-cli"]
+members = ["crates/rullm-cli", "crates/rullm-chat-completion", "crates/rullm-anthropic"]
 resolver = "2"
 
 [workspace.package]
@@ -15,7 +15,7 @@ rand = "0.8"
 reqwest = { version = "0.11", features = ["json", "stream"] }
 bytes = "1.0"
 log = "0.4"
-serde = { version = "1.0", features = ["derive"] }
+serde = { version = "1.0", features = ["derive", "rc"] }
 serde_json = "1.0"
 thiserror = "1.0"
 anyhow = "1.0"

README.md

@@ -13,7 +13,6 @@ rullm "What is the capital of France?"
 # Use different models with aliases
 rullm --model gpt4 "Explain quantum computing"
 rullm --model claude "Write a poem about the ocean"
-rullm --model gemini "What's the weather like?"
 
 # Use templates for structured queries ({{input}} placeholder is automatically filled)
 rullm -t code-review "Review this function"
@@ -32,7 +31,7 @@ rullm chat --model claude
 
 # Disable streaming for buffered output
 rullm --no-streaming "Write a poem about the ocean"
-rullm chat --no-streaming --model gemini
+rullm chat --no-streaming --model claude
 
 # Set up your API keys
 rullm keys set openai

crates/rullm-anthropic/Cargo.toml

@@ -0,0 +1,18 @@
+[package]
+name = "rullm-anthropic"
+version.workspace = true
+edition.workspace = true
+rust-version.workspace = true
+
+[dependencies]
+tokio = { workspace = true }
+reqwest = { workspace = true }
+bytes = { workspace = true }
+serde = { workspace = true, features = ["rc"] }
+serde_json = { workspace = true }
+thiserror = { workspace = true }
+futures = { workspace = true }
+async-stream = { workspace = true }
+
+[dev-dependencies]
+tokio-test = { workspace = true }