@simulacra-ai/anthropic 0.0.2 → 0.0.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (2) hide show
  1. package/README.md +9 -6
  2. package/package.json +2 -2
package/README.md CHANGED
@@ -1,6 +1,6 @@
1
1
  # Simulacra Anthropic Provider
2
2
 
3
- Anthropic Claude provider for the Simulacra conversation engine.
3
+ The Anthropic provider allows Simulacra to use Claude models via the Anthropic API, with support for extended thinking and prompt caching.
4
4
 
5
5
  ## Installation
6
6
 
@@ -15,8 +15,9 @@ import { Conversation } from "@simulacra-ai/core";
15
15
  import { AnthropicProvider } from "@simulacra-ai/anthropic";
16
16
  import Anthropic from "@anthropic-ai/sdk";
17
17
 
18
+ // create a provider and conversation
18
19
  const provider = new AnthropicProvider(new Anthropic(), { model: MODEL_NAME });
19
- const conversation = new Conversation(provider);
20
+ using conversation = new Conversation(provider);
20
21
  ```
21
22
 
22
23
  ### AnthropicProviderConfig
@@ -32,9 +33,11 @@ interface AnthropicProviderConfig {
32
33
 
33
34
  Additional properties (`temperature`, `top_p`, etc.) spread into the API request.
34
35
 
36
+ If `max_tokens` is not set, it defaults to 8192.
37
+
35
38
  ## Extended Thinking
36
39
 
37
- Enable extended thinking to capture the model's chain-of-thought as `ThinkingMessageContent` blocks.
40
+ Extended thinking captures the model's chain-of-thought reasoning as `ThinkingMessageContent` blocks in the conversation history. This makes the model's reasoning process visible and accessible alongside its regular output.
38
41
 
39
42
  ```typescript
40
43
  const provider = new AnthropicProvider(new Anthropic(), {
@@ -44,11 +47,11 @@ const provider = new AnthropicProvider(new Anthropic(), {
44
47
  });
45
48
  ```
46
49
 
47
- Both `thinking` and `redacted_thinking` blocks from the API are converted to Simulacra's `ThinkingMessageContent` type. If `max_tokens` is not set, it defaults to 8192.
50
+ Both `thinking` and `redacted_thinking` blocks from the API are converted to Simulacra's `ThinkingMessageContent` type.
48
51
 
49
52
  ## Prompt Caching
50
53
 
51
- Prompt caching is enabled by default. The provider adds `cache_control` markers to the system prompt and tool definitions, avoiding re-processing static content on every turn.
54
+ Prompt caching is enabled by default. The provider adds `cache_control` markers to the system prompt and tool definitions, avoiding re-processing static content on every turn. Multi-turn conversations benefit because subsequent turns read from the cache at a reduced rate.
52
55
 
53
56
  ```typescript
54
57
  const provider = new AnthropicProvider(new Anthropic(), {
@@ -61,7 +64,7 @@ const provider = new AnthropicProvider(new Anthropic(), {
61
64
  });
62
65
  ```
63
66
 
64
- Cache writes cost slightly more per token than standard requests. Multi-turn conversations benefit because subsequent turns read from the cache at a reduced rate. Disable by setting both flags to `false`.
67
+ Cache writes cost slightly more per token than standard requests. Caching can be disabled by setting both flags to `false`.
65
68
 
66
69
  ## License
67
70
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@simulacra-ai/anthropic",
3
- "version": "0.0.2",
3
+ "version": "0.0.3",
4
4
  "description": "Anthropic Claude provider for the Simulacra conversation engine",
5
5
  "type": "module",
6
6
  "exports": {
@@ -17,7 +17,7 @@
17
17
  "clean": "rm -rf dist *.tsbuildinfo"
18
18
  },
19
19
  "peerDependencies": {
20
- "@simulacra-ai/core": "0.0.2",
20
+ "@simulacra-ai/core": "0.0.3",
21
21
  "@anthropic-ai/sdk": ">=0.30.0 <1.0.0"
22
22
  },
23
23
  "repository": {