Adds `AssistantMessage.responseModel` on the openai-completions path:
surfaces the concrete `chunk.model` when it differs from the requested
id (e.g. OpenRouter `auto` -> `anthropic/...`).
* feat(ai): add Cloudflare Workers AI as a provider
Cloudflare Workers AI hosts open-weight LLMs (Kimi K2.6, GPT-OSS,
GLM-4.7, Llama 4, Gemma 4, Nemotron 3) on Cloudflare's GPU network with
an OpenAI-compatible endpoint. Reuses the openai-completions API
protocol; the per-account URL contains a {CLOUDFLARE_ACCOUNT_ID}
placeholder resolved at request time by a small helper.
Pi automatically sets x-session-affinity for prefix caching:
https://developers.cloudflare.com/workers-ai/features/prompt-caching/
Auth: CLOUDFLARE_API_KEY (matches pi's *_API_KEY convention) +
CLOUDFLARE_ACCOUNT_ID. The User-Agent identifies traffic as
'pi-coding-agent' in Cloudflare analytics.
Verified end-to-end against a real Cloudflare account: 17 e2e tests
pass across stream/empty/tokens/unicode/tool-call-without-result/
total-tokens against @cf/moonshotai/kimi-k2.6.
Cloudflare AI Gateway is a separate, larger change (it requires routing
through provider-specific subpaths with the matching API protocol per
upstream) and will land in a follow-up PR.
* refactor(ai): move Cloudflare User-Agent and session-affinity flag to per-model metadata
Instead of conditionally setting them in openai-completions.ts based on
provider detection, declare them as model-level fields in the catalog
(headers + compat). This is consistent with how the github-copilot and
kimi-coding entries already declare their static headers.
packages/ai/scripts/generate-models.ts: emit headers and compat fields
on each cloudflare-workers-ai entry (CLOUDFLARE_STATIC_HEADERS).
packages/ai/src/providers/openai-completions.ts: drop the
isCloudflareProvider conditional that injected User-Agent and the
isCloudflareWorkersAI override of sendSessionAffinityHeaders.
packages/ai/src/models.generated.ts: re-spliced 8 cloudflare-workers-ai
entries with headers + compat.
Behavior is unchanged - verified via fetch interceptor that User-Agent
and x-session-affinity / session_id / x-client-request-id are still sent
on outbound requests. 5/5 e2e tests pass.
supportsPromptCaching, supportsAdaptiveThinking, supportsThinkingSignature,
and the Claude detection in streamSimpleBedrock/buildAdditionalModelRequestFields
all check model.id for Claude model name patterns. Application inference profile
ARNs are opaque and do not contain the model name, so these checks silently fail.
Fix by also checking model.name (user-controlled via models.json or
registerProvider) as a fallback in all affected functions. Added a shared
isAnthropicClaudeModel helper for the common Claude detection pattern.
Fixes#2925
Co-authored-by: Your Name <you@example.com>
DashScope / Aliyun Qwen (OpenAI-compatible) rejects `tools: []`
with HTTP 400 `"[] is too short - 'tools'"`. Five providers used
a truthy check (`if (context.tools)`) that treated an empty array
as "send tools", so `pi --no-tools` produced `tools: []` in the
request body. Matching the Google provider's pattern, we now
guard on `context.tools.length > 0`:
- openai-completions.ts
- openai-responses.ts
- openai-codex-responses.ts
- azure-openai-responses.ts
- anthropic.ts
The openai-completions fallback that emits `tools: []` when the
conversation has tool history (required by LiteLLM / Anthropic
proxies) is preserved via the existing `else if (hasToolHistory)`
branch.
closes#3649
Co-authored-by: 槐聚 <huaiju@zbyte-inc.com>
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: Mario Zechner <badlogicgames@gmail.com>
* fix(typebox): migrate to v1 with extension compat
Replace AJV-based validation with TypeBox-native validation, keep legacy extension imports working (including @sinclair/typebox/compiler), and restore coercion for serialized/plain JSON schemas.
This change closes#3112.
* fix(typebox): use canonical imports and harden coercion
Switch first-party code to canonical typebox imports while retaining legacy extension aliases in the loader.
Remove obsolete runtime codegen guards, expand serialized JSON-schema coercion coverage, and update related tests and fixtures.
Fixes#3112.
---------
Co-authored-by: Mario Zechner <badlogicgames@gmail.com>
Replace the deprecated global anthropic-beta header with per-tool
eager_input_streaming: true. Fix empty anthropic-beta header causing
400 errors on models with no beta features.
fixes#3175
Bypass the Anthropic SDK streaming parser entirely. Use
client.messages.create().asResponse() and decode the SSE stream
ourselves with defensive JSON parsing that repairs invalid escape
sequences and control characters inside string literals.
- Switch from SDK .stream() to .asResponse() + pi-owned SSE decoder
- Add repairJson() / parseJsonWithRepair() to json-parse.ts
- Add anthropic-sse-parsing.test.ts regression for malformed tool deltas
- Update github-copilot-anthropic.test.ts mock to match new call path
- Update deprecated claude-3-5-haiku-20241022 refs to claude-haiku-4-5
- Remove stale non-reasoning model test
fixes#3175
When model.maxTokens is 0 (unset/unknown), omit maxTokens from the
Bedrock ConverseStream inferenceConfig instead of sending 0.
Bedrock's inferenceConfig.maxTokens is optional — when omitted, the
model uses its own internal default. This is optimal for Bedrock's
token quota management:
- At request start, Bedrock reserves input_tokens + max_tokens from
your TPM quota
- For Claude 3.7+ models, output tokens have a 5x burndown rate
- Sending an unnecessarily high maxTokens wastes TPM capacity during
the reservation window, reducing concurrent request throughput
- Omitting it lets Bedrock use the model default (~4096 for Claude),
matching expected output sizes and maximizing quota utilization
Also applies the same conditional pattern to temperature — only
include it in inferenceConfig when explicitly set.
Changes:
- simple-options.ts: Use ?? instead of || so explicit 0 is preserved;
return undefined when model.maxTokens is 0 (unset)
- amazon-bedrock.ts: Spread maxTokens and temperature into
inferenceConfig only when defined
Fixes#3399
BedrockRuntimeClient was constructed without the endpoint option,
causing model.baseUrl to be silently ignored. This meant custom
Bedrock endpoints (VPC endpoints, proxy setups, custom routing)
were never used — requests always went to the default regional
endpoint.
Pass model.baseUrl as the endpoint config when set. When not set
(the default), the SDK falls back to the standard regional endpoint
constructed from the region config, preserving existing behavior.
Fixesopenclaw/openclaw#47899
Adds bearer token authentication for the Bedrock Converse API, enabling
users to authenticate with an API key instead of SigV4/IAM credentials.
When a bearer token is available (via `options.bearerToken` or the
`AWS_BEARER_TOKEN_BEDROCK` env var), the provider:
1. Sets dummy credentials to prevent SDK credential resolution errors
2. Injects middleware after SigV4 signing that replaces the Authorization
header with `Bearer <token>` and removes SigV4-specific headers
This uses the official `bedrock:CallWithBearerToken` IAM action, which
is a documented AWS feature for API key auth on Bedrock endpoints.
Use case: users who receive a Bedrock API key (bearer token) from the
AWS console or their admin, without having IAM access keys or instance
roles. Similar to how ANTHROPIC_API_KEY works for direct Anthropic API.
Required IAM permission on the token's identity:
bedrock:CallWithBearerToken
Tested: Bearer token successfully authenticates against Bedrock Converse
API (returns correct 403 for missing IAM permission, not auth format error).
SigV4 path is unchanged when no bearer token is set.
Exposes the new ThinkingConfig.display field on Anthropic and Bedrock
Claude providers. Defaults to 'summarized' so Claude Opus 4.7 and Mythos
Preview keep returning thinking text despite Anthropic's silent default
change to 'omitted'. Set to 'omitted' explicitly to skip thinking
streaming for faster time-to-first-text-token.
Mario Zechner