mirror of
https://github.com/whit3rabbit/anyllm-proxy.git
synced 2026-09-22 16:00:51 +00:00
whit3rabbitandClaude Opus 4.6
f2e3ed15f4
Rust workspace with two crates: - translator: pure, IO-free mapping between Anthropic Messages API and OpenAI Chat Completions - proxy: axum HTTP server with auth, streaming SSE, retry/backoff, concurrency limits 169 tests passing (unit, golden fixture, integration). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
65 lines
2.3 KiB
Markdown
65 lines
2.3 KiB
Markdown
# Anthropic API: Type Verification Notes
|
|
|
|
## Verified Against
|
|
|
|
Anthropic Messages API documentation as referenced in PLAN.md.
|
|
|
|
## MessageCreateRequest
|
|
|
|
| Field | Type | Required | Implemented | Notes |
|
|
|---|---|---|---|---|
|
|
| model | string | Yes | Yes | |
|
|
| max_tokens | u32 | Yes | Yes | |
|
|
| messages | Vec<InputMessage> | Yes | Yes | |
|
|
| system | Option<System> | No | Yes | String or blocks |
|
|
| temperature | Option<f32> | No | Yes | Range 0..1 |
|
|
| top_p | Option<f32> | No | Yes | |
|
|
| stop_sequences | Option<Vec<String>> | No | Yes | |
|
|
| tools | Option<Vec<Tool>> | No | Yes | |
|
|
| tool_choice | Option<ToolChoice> | No | Yes | auto/any/none/tool |
|
|
| metadata | Option<Metadata> | No | Yes | user_id only |
|
|
| stream | Option<bool> | No | Yes | |
|
|
| thinking | - | No | Not implemented | Extended thinking config |
|
|
|
|
## Content Blocks
|
|
|
|
| Block Type | Implemented | Notes |
|
|
|---|---|---|
|
|
| text | Yes | |
|
|
| image | Yes | base64 and URL sources |
|
|
| document | Yes | Converted to text note (no inline PDF in Chat Completions) |
|
|
| tool_use | Yes | id, name, input (JSON object) |
|
|
| tool_result | Yes | tool_use_id, content (string or blocks), is_error |
|
|
| thinking | No | Extended thinking not proxied |
|
|
|
|
## MessageResponse
|
|
|
|
| Field | Implemented | Notes |
|
|
|---|---|---|
|
|
| id | Yes | Generated as msg_{uuid} |
|
|
| type | Yes | Always "message" |
|
|
| role | Yes | Always "assistant" |
|
|
| content | Yes | Vec<ContentBlock> |
|
|
| model | Yes | Echoed from request |
|
|
| stop_reason | Yes | end_turn, max_tokens, stop_sequence, tool_use |
|
|
| stop_sequence | Yes | Always null (not tracked from OpenAI) |
|
|
| usage | Yes | input_tokens, output_tokens |
|
|
|
|
## Error Types
|
|
|
|
All 8 error types implemented: invalid_request_error, authentication_error, permission_error, not_found_error, request_too_large, rate_limit_error, api_error, overloaded_error.
|
|
|
|
## Streaming Events
|
|
|
|
All event types implemented: message_start, content_block_start, content_block_delta, content_block_stop, message_delta, message_stop, ping, error.
|
|
|
|
Delta types: text_delta, input_json_delta. Thinking_delta not implemented.
|
|
|
|
## Known Gaps
|
|
|
|
- Extended thinking (`thinking` config and `thinking_delta` events)
|
|
- `pause_turn` stop reason (no OpenAI equivalent)
|
|
- Prompt caching (`cache_control` on system blocks parsed but not utilized)
|
|
- Beta Files API
|
|
- Beta Skills API
|