Models with supportsReasoningEffort as an array (e.g., gpt-5.2) and no
requiredReasoningEffort would show a blank dropdown because the default
fell to 'disable' which wasn't in the available options. Now uses the
model's declared reasoningEffort as the default.
The API version field is an advanced option that 99% of users won't change.
Hide it in the onboarding Welcome view using the existing simplifySettings
pattern (matches OpenRouter's approach for custom base URL).
Pasted URLs often contain older API versions (e.g., 2024-05-01-preview) that
are incompatible with the Responses API. The parser now only extracts endpoint
and deployment name, letting the API version default to 2025-04-01-preview.
- Fix baseURL empty string fallback (pass undefined instead of "" to let SDK use env vars)
- Add parseAzureUrl() utility that extracts endpoint, deployment name, and API version from a full Azure deployment URL
- Integrate auto-parser into Azure settings: pasting a full URL auto-fills all fields
- Rename "Base URL" to "Azure Endpoint" to match Azure portal terminology
- Improve field descriptions for deployment name and API version
- Add 13 tests for URL parser (all passing)
Expand azure model list from 27 (OpenAI-only) to 64 models with
tool_call support, including Claude, Cohere, DeepSeek, Grok, Kimi,
Llama, Mistral, Phi, and Model Router families.
All model metadata (pricing, context window, capabilities) sourced
from https://models.dev/api.json. Existing Roo-specific overrides
(includedTools, excludedTools, reasoningEffort, etc.) preserved.
- Upgrade @ai-sdk/azure from ^2.0.6 to ^3.0.26
- v3 defaults provider(id) to Responses API (works for all models)
- v2 defaulted to Chat Completions, breaking codex models
- Add useDeploymentBasedUrls: true to createAzure()
- Produces /deployments/{id}/{path} URLs (universally compatible)
- Without it, SDK uses /v1/{path} which only works on AI Foundry resources
- Add static model metadata for 29 Azure models (from models.dev)
with Roo-specific flags (reasoning, tools, verbosity) matching
openAiNativeModels
- Add model picker dropdown to Azure provider settings for model
capability detection (context window, max tokens, pricing)
- Rename provider label from 'Azure OpenAI' to 'Azure AI Foundry'
across all 18 locales
- Make API key optional (supports Azure managed identity / Entra ID)
- Update default API version from 2024-08-01-preview to 2025-04-01-preview
- Fix maxOutputTokens validation (filter invalid values <= 0)
- Handler separates deployment name (API calls) from model ID
(capability lookup) with azureDefaultModelInfo (gpt-4o) fallback
- Remove unhelpful 'Get Azure AI Foundry Access' button
- Prevent stale model IDs from other providers carrying over
- Suppress validation errors on fresh provider selection
- Add azureApiKey to SECRET_STATE_KEYS for proper configuration detection
- Add Azure validation case in validateModelsAndKeysProvided
- Add validation translations for azureResourceName and azureDeploymentName across all 18 locales
This fixes the issue where the Finish button does nothing when setting up Azure provider in the onboarding workflow.
* refactor: migrate baseten provider to AI SDK
* refactor(baseten): migrate to native @ai-sdk/baseten package
Replace OpenAICompatibleHandler with dedicated @ai-sdk/baseten package,
following the same pattern used by other native AI SDK providers (groq,
deepseek, etc.). This uses createBaseten() for provider initialization
and extends BaseProvider directly instead of the generic OpenAI-compatible
handler.
Add GetCommands, GetModes, and GetModels to the IPC protocol so external
clients can fetch slash commands, available modes, and Roo provider models
without going through the internal webview message channel.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* feat: add disabledTools setting to globally disable native tools
Add a disabledTools field to GlobalSettings that allows disabling specific
native tools by name. This enables cloud agents to be configured with
restricted tool access.
Schema:
- Add disabledTools: z.array(toolNamesSchema).optional() to globalSettingsSchema
- Add disabledTools to organizationDefaultSettingsSchema.pick()
- Add disabledTools to ExtensionState Pick type
Prompt generation (tool filtering):
- Add disabledTools to BuildToolsOptions interface
- Pass disabledTools through filterSettings to filterNativeToolsForMode()
- Remove disabled tools from allowedToolNames set in filterNativeToolsForMode()
Execution-time validation (safety net):
- Extract disabledTools from state in presentAssistantMessage
- Convert disabledTools to toolRequirements format for validateToolUse()
Wiring:
- Add disabledTools to ClineProvider getState() and getStateToPostToWebview()
- Pass disabledTools to all buildNativeToolsArrayWithRestrictions() call sites
EXT-778
* fix: check toolRequirements before ALWAYS_AVAILABLE_TOOLS
Moves the toolRequirements check before the ALWAYS_AVAILABLE_TOOLS
early-return in isToolAllowedForMode(). This ensures disabledTools
can block always-available tools (switch_mode, new_task, etc.) at
execution time, making the validation layer consistent with the
filtering layer.
* feat: migrate Bedrock provider to AI SDK
Replace the raw AWS SDK (@aws-sdk/client-bedrock-runtime) Bedrock handler
with the Vercel AI SDK (@ai-sdk/amazon-bedrock). Reduces provider from
1,633 lines to 575 lines (65% reduction).
Key changes:
- Use streamText()/generateText() instead of ConverseStreamCommand/ConverseCommand
- Use createAmazonBedrock() with native auth (access key, secret, session,
profile via credentialProvider, API key, VPC endpoint as baseURL)
- Reasoning config via providerOptions.bedrock.reasoningConfig
- Anthropic beta headers via providerOptions.bedrock.anthropicBeta
- Thinking signature captured from providerMetadata.bedrock.signature
on reasoning-delta stream events
- Thinking signature round-tripped via providerOptions.bedrock.signature
on reasoning parts in convertToAiSdkMessages()
- Redacted thinking captured from providerMetadata.bedrock.redactedData
- isAiSdkProvider() returns true for reasoning block preservation
- Keep: getModel, ARN parsing, cross-region inference, cost calculation,
service tier pricing, 1M context beta
Tests: 83 tests skipped (mock old AWS SDK internals, need rewrite for
AI SDK mocking). 106 tests pass. 0 tests fail.
* fix: address review feedback for Bedrock AI SDK migration
- Wire usePromptCache into AI SDK via providerOptions.bedrock.cachePoint
on system prompt and last two user messages
- Remove debug logger.info that fires on every stream event with
providerMetadata
- Tighten isThrottlingError to match 'rate limit' instead of broad
'rate'/'limit' substrings that false-positive on context length errors
- Use shared handleAiSdkError utility for consistent error handling
with status code preservation for retry logic
* fix: bedrock AI SDK migration - fix usage metrics, rewrite tests, remove dead code
- Fix reasoningTokens always 0 (usage.details?.reasoningTokens → usage.reasoningTokens)
- Fix cacheReadInputTokens always 0 (read from usage.inputTokenDetails instead of providerMetadata)
- Fix invokedModelId not extracted for prompt router cost calculation
- Rewrite all 6 skipped bedrock test suites for AI SDK mocking pattern (140 tests pass)
- Remove dead code: bedrock-converse-format.ts, cache-strategy/ (6 files, ~2700 lines)
* chore: remove dead @anthropic-ai/bedrock-sdk dep and stale AWS SDK mocks
* chore: update pnpm-lock.yaml after removing @anthropic-ai/bedrock-sdk
* fix: compute cache point indices from original Anthropic messages before AI SDK conversion
The previous approach naively targeted the last 2 user messages in the
post-conversion AI SDK array, but convertToAiSdkMessages() splits user
messages containing tool_results into separate tool + user messages,
causing cache points to land on the wrong messages (tiny text fragments
instead of the intended meaty user turns).
Now we identify the last 2 user messages in the original Anthropic
message array (matching the Anthropic provider's caching strategy) and
build a parallel-walk mapping to apply cachePoint to the correct
corresponding AI SDK message.
* perf: optimize prompt caching with 3-point message strategy + anchor for 20-block window
Previous approach only cached the last 2 user messages (using 2 of 4
available cache checkpoints for messages). This left significant cache
savings on the table for longer conversations.
New strategy uses up to 3 message cache points (+ 1 system = 4 total):
- Last user message: write to cache for next request
- Second-to-last user message: read from cache for current request
- Anchor message at ~1/3 position: ensures the 20-block lookback window
from the second-to-last breakpoint hits a stable cache entry, covering
all assistant/tool messages in the middle of the conversation
Also extracted the parallel-walk mapping logic into a reusable
applyCachePointsToAiSdkMessages() helper method.
Industry benchmarks show 70-95% token cache rates are achievable;
this change should significantly improve our 39% baseline for longer
multi-turn conversations.
* chore: remove stale bedrock-sdk external, fix arnInfo property name, remove unused exports
---------
Co-authored-by: daniel-lxs <ricciodaniel98@gmail.com>
* fix: DeepSeek temperature defaulting to 0 instead of 0.3
Pass defaultTemperature: DEEP_SEEK_DEFAULT_TEMPERATURE to getModelParams() in
DeepSeekHandler.getModel() to ensure the correct default temperature (0.3)
is used when no user configuration is provided.
Closes#11194
* refactor: make defaultTemperature required in getModelParams
Make the defaultTemperature parameter required in getModelParams() instead
of defaulting to 0. This prevents providers with their own non-zero default
temperature (like DeepSeek's 0.3) from being silently overridden by the
implicit 0 default.
Every provider now explicitly declares its temperature default, making the
temperature resolution chain clear:
user setting → model default → provider default
---------
Co-authored-by: Roo Code <roomote@roocode.com>
Co-authored-by: daniel-lxs <ricciodaniel98@gmail.com>
* refactor: migrate featherless provider to AI SDK
* fix: merge consecutive same-role messages in featherless R1 path
convertToAiSdkMessages does not merge consecutive same-role messages
like convertToR1Format did. When the system prompt is prepended as a
user message and the conversation already starts with a user message,
DeepSeek R1 can reject the request.
Add mergeConsecutiveSameRoleMessages helper that collapses adjacent
Anthropic messages sharing the same role before AI SDK conversion.
Includes a test that verifies no two successive messages share a role.
Migrates the IO Intelligence provider from legacy BaseOpenAiCompatibleProvider
(direct openai SDK) to OpenAICompatibleHandler (Vercel AI SDK).
- Extends OpenAICompatibleHandler instead of BaseOpenAiCompatibleProvider
- Uses getModelParams for model parameter resolution
- Updates tests to mock ai module's streamText/generateText
Remove two console.warn messages that fire excessively when loading tasks
from history:
- 'Attempting to finalize unknown tool call' in finalizeStreamingToolCall()
- 'Received chunk for unknown tool call' in processStreamingChunk()
The defensive null-return behavior is preserved; only the log output is removed.
* changeset version bump
* Update CHANGELOG for version 3.47.3
Updated version number and removed redundant patch changes for 3.47.3.
---------
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Matt Rubens <mrubens@users.noreply.github.com>
Remove the "Enable URL context" and "Enable Grounding with Google search"
checkboxes from Gemini and Vertex provider settings, along with:
- enableUrlContext and enableGrounding fields from provider settings schemas
- URL context and Google Search tool injection in completePrompt methods
- Associated translation keys from all 18 locale files
- Related test cases updated to reflect the removal
- simplifySettings prop removed from Gemini and Vertex components
(it was only used for the removed checkboxes in those components)
Co-authored-by: Roo Code <roomote@roocode.com>
* changeset version bump
* Update CHANGELOG for version 3.47.2
Updated version number and added patch changes for 3.47.2.
* Update CHANGELOG for version 3.47.1
Updated changelog for version 3.47.1 with fixes and cleanup.
---------
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Matt Rubens <mrubens@users.noreply.github.com>
* fix: capture and round-trip thinking signature for Bedrock Claude models
Bedrock handler streams reasoning text from Claude's extended thinking but
never captures the cryptographic signature. This causes 400 errors on
multi-turn conversations with tool use: 'Expected thinking or
redacted_thinking, but found tool_use'.
Changes:
- bedrock.ts: Capture reasoningContent.signature from Converse API stream
deltas, implement getThoughtSignature() so Task.ts stores it as a proper
thinking content block
- bedrock-converse-format.ts: Convert thinking blocks to Bedrock's
reasoningContent format with signature, skip reasoning/redacted_thinking/
thoughtSignature blocks that aren't valid for the API
* fix: add redacted_thinking round-trip, fix interface types, add tests
Address PR review feedback:
- Update ContentBlockDeltaEvent interface to include signature and
redactedContent fields (removes type assertions)
- Add 6 tests for thinking/reasoning block conversions in
bedrock-converse-format.ts
Also add redacted_thinking round-trip support:
- bedrock.ts: Capture redactedContent from stream deltas, base64 encode,
expose via getRedactedThinkingBlocks()
- Task.ts: Insert redacted_thinking blocks after thinking block in
assistant messages
- bedrock-converse-format.ts: Convert redacted_thinking blocks back to
reasoningContent.redactedContent (base64 → Uint8Array)
PR #11180 migrated Gemini/Vertex providers to the AI SDK and deleted
gemini-format.ts which contained the working thought signature round-trip
logic (originally added in PR #10590). This broke all Gemini 3 tool use
with a 400 error: 'Function call is missing a thought_signature'.
Changes:
- Gemini/Vertex handlers: capture thoughtSignature from providerMetadata
on tool-call stream events, expose via getThoughtSignature()
- convertToAiSdkMessages(): extract thoughtSignature content blocks from
history, attach as providerOptions on first tool-call part (per Gemini 3
parallel call rules)
- Add 3 tests verifying thought signature round-trip behavior
* feat: add support for .agents/skills directory
This change adds support for discovering skills from the .agents/skills
directory, following the Agent Skills convention for sharing skills
across different AI coding tools.
Priority order (later entries override earlier ones):
1. Global ~/.agents/skills (shared across AI coding tools, lowest priority)
2. Project .agents/skills
3. Global ~/.roo/skills (Roo-specific)
4. Project .roo/skills (highest priority)
Changes:
- Add getGlobalAgentsDirectory() and getProjectAgentsDirectoryForCwd()
functions to roo-config
- Update SkillsManager.getSkillsDirectories() to include .agents/skills
- Update SkillsManager.setupFileWatchers() to watch .agents/skills
- Add tests for new functionality
* fix: clarify skill priority comment to match actual behavior
* fix: clarify skill priority comment to explain Map.set replacement mechanism
---------
Co-authored-by: Roo Code <roomote@roocode.com>
* changeset version bump
* Update CHANGELOG for version 3.47.1
Updated changelog for version 3.47.1 with patch changes.
---------
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: John Richmond <5629+jr@users.noreply.github.com>
When the 'custom base URL' checkbox is unchecked in the UI, the setting
is set to '' (empty string). Providers that passed this directly to their
SDK constructors caused 'Failed to parse URL' errors because the SDK
treated '' as a valid but broken base URL override.
- gemini.ts: use || undefined (was passing raw option)
- openai-native.ts: use || undefined (was passing raw option)
- openai.ts: change ?? to || for fallback default
- deepseek.ts: change ?? to || for fallback default
- moonshot.ts: change ?? to || for fallback default
Adds test coverage for Gemini and OpenAI Native constructors verifying
empty-string baseURL is coerced to undefined.
Remove the :0 suffix from the Claude Opus 4.6 model ID to match
the correct AWS Bedrock model identifier.
The model ID was "anthropic.claude-opus-4-6-v1:0" but should be
"anthropic.claude-opus-4-6-v1" per AWS Bedrock documentation.
Fixes#11231
Co-authored-by: Roo Code <roomote@roocode.com>
Remove three functions from appendEnvironmentDetails.ts that were
defined and tested but never imported or called in production code:
- stripAppendedEnvironmentDetails (exported, 0 call sites)
- stripEnvDetailsFromText (private helper)
- stripEnvDetailsFromToolResult (private helper)
Also removes the corresponding describe block (7 tests) from the
spec file. The remaining 19 tests pass.
