romtuck/openclaw
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: add pnpm patch for pi-ai to support LiteLLM providerType

Browse Source 
When using Claude models via LiteLLM proxy, the bundled pi-ai sends
OpenAI-specific parameters (store, max_completion_tokens) that cause
errors with non-OpenAI backends like AWS Bedrock.

This patch adds providerType support to openai-completions.js:
- Check model.providerType to determine if model is OpenAI-native
- Only send OpenAI-specific params to actual OpenAI models
- Use max_tokens instead of max_completion_tokens for proxy models

Users can now configure custom models in ~/.pi/agent/models.json with
providerType: "anthropic" (or "google", etc.) to properly route
requests through LiteLLM to Claude and other non-OpenAI models.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Ubuntu 2025-12-08 01:07:57 +00:00
parent 2112fa919a
commit 3da4d91ec0
2 changed files with 43 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
patches/@mariozechner__pi-ai.patch Normal file
View File

@ -0,0 +1,40 @@
diff --git a/dist/providers/openai-completions.js b/dist/providers/openai-completions.js
index a430c2240a37bcc7d05fb8e551c18964e7636c59..869eb60f0724c31b08aba2e2ee37da63d82696b2 100644
--- a/dist/providers/openai-completions.js
+++ b/dist/providers/openai-completions.js
@@ -231,20 +231,26 @@ function buildParams(model, context, options) {
stream: true,
stream_options: { include_usage: true },
};
- // Cerebras/xAI/Mistral dont like the "store" field
- if (!model.baseUrl.includes("cerebras.ai") &&
- !model.baseUrl.includes("api.x.ai") &&
- !model.baseUrl.includes("mistral.ai") &&
- !model.baseUrl.includes("chutes.ai")) {
+ // Only send OpenAI-specific params (like "store") to actual OpenAI models.
+ // For custom models via proxies like LiteLLM, check providerType.
+ // Fall back to baseUrl detection for known providers.
+ const isOpenAIModel = model.providerType === "openai" ||
+ (!model.providerType &&
+ !model.baseUrl.includes("cerebras.ai") &&
+ !model.baseUrl.includes("api.x.ai") &&
+ !model.baseUrl.includes("mistral.ai") &&
+ !model.baseUrl.includes("chutes.ai") &&
+ (model.baseUrl.includes("api.openai.com") || model.provider === "openai"));
+ if (isOpenAIModel) {
params.store = false;
}
if (options?.maxTokens) {
- // Mistral/Chutes uses max_tokens instead of max_completion_tokens
- if (model.baseUrl.includes("mistral.ai") || model.baseUrl.includes("chutes.ai")) {
- params.max_tokens = options?.maxTokens;
+ // Only OpenAI uses max_completion_tokens. Others (Mistral/Chutes/Anthropic via proxy) use max_tokens
+ if (isOpenAIModel) {
+ params.max_completion_tokens = options?.maxTokens;
}
else {
- params.max_completion_tokens = options?.maxTokens;
+ params.max_tokens = options?.maxTokens;
}
}
if (options?.temperature !== undefined) {

3
pnpm-workspace.yaml
View File

@ -2,3 +2,6 @@ onlyBuiltDependencies:
- '@whiskeysockets/baileys'
- esbuild
- sharp
patchedDependencies:
'@mariozechner/pi-ai': patches/@mariozechner__pi-ai.patch