Escape potential XML/HTML-like tags in untrusted tool outputs
(command output, file contents, binary extraction) before they are
fed back into the LLM context. This mitigates indirect prompt
injection via malicious file contents or shell output.
- Add sanitizeForPromptInjection() helper to text-normalization
- Apply sanitization in ExecuteCommandTool result formatting
- Apply sanitization in ReadFileTool text and binary paths
- Apply sanitization in extract-text binary extractors
* fix: add image content support to MCP tool responses (#10874)
Co-authored-by: Roo Code <roomote@roocode.com>
* fix: transform tool blocks to text before condensing (EXT-624) (#10975)
* refactor(read_file): Codex-inspired read_file refactor EXT-617 (#10981)
* feat: allow import settings in initial welcome screen (#10994)
Co-authored-by: Roo Code <roomote@roocode.com>
Co-authored-by: Matt Rubens <mrubens@users.noreply.github.com>
* fix(code-index): remove deprecated text-embedding-004 and migrate to gemini-embedding-001 (#11038)
Co-authored-by: Roo Code <roomote@roocode.com>
Co-authored-by: Hannes Rudolph <hrudolph@gmail.com>
* chore: treat extension .env as optional (#11116)
* fix: sanitize tool_use_id in tool_result blocks to match API history (#11131)
Tool IDs from providers like Gemini/OpenRouter contain special characters
(e.g., 'functions.read_file:0') that are sanitized when saving tool_use
blocks to API history. However, tool_result blocks were using the original
unsanitized IDs, causing ToolResultIdMismatchError.
This fix ensures tool_result blocks use sanitizeToolUseId() to match the
sanitized tool_use IDs in conversation history.
Fixes EXT-711
* fix: queue messages during command execution instead of losing them (#11140)
* IPC fixes for task cancellation and queued messages (#11162)
* feat: add support for AGENTS.local.md personal override files (#11183)
Co-authored-by: Roo Code <roomote@roocode.com>
Co-authored-by: roomote[bot] <219738659+roomote[bot]@users.noreply.github.com>
* fix(cli): resolve race condition causing provider switch during mode changes (#11205)
When using slash commands with `mode:` frontmatter (e.g., `/cli-release`
with `mode: code`), the CLI would fail with "Could not resolve
authentication method" from the Anthropic SDK, even when using a
non-Anthropic provider like `--provider roo`.
Root cause: In `markWebviewReady()`, the `webviewDidLaunch` message was
sent before `updateSettings`, creating a race condition. The
`webviewDidLaunch` handler's "first-time init" sync would read
`getState()` before CLI-provided settings were applied to the context
proxy. Since `getState()` defaults `apiProvider` to "anthropic" when
unset, this default was saved to the provider profile. When a slash
command triggered `handleModeSwitch()`, it found this corrupted profile
with `apiProvider: "anthropic"` (but no API key) and activated it,
overwriting the CLI's working roo provider configuration.
Fix:
1. Reorder `markWebviewReady()` to send `updateSettings` before
`webviewDidLaunch`, ensuring the context proxy has CLI-provided
values when the initialization handler runs.
2. Guard the first-time init sync with `checkExistKey(apiConfiguration)`
to prevent saving a profile with only the default "anthropic"
fallback and no actual API keys configured.
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
* chore: remove dead toolFormat code from getEnvironmentDetails (#11207)
Remove the toolFormat constant and <tool_format> line from environment
details output. Native tool calling is now the only supported protocol,
making this code unnecessary.
Fixes#11206
Co-authored-by: Roo Code <roomote@roocode.com>
* feat: extract translation and merge resolver modes into reusable skills (#11215)
* feat: extract translation and merge resolver modes into reusable skills
- Add roo-translation skill with comprehensive i18n guidelines
- Add roo-conflict-resolution skill for intelligent merge conflict resolution
- Add /roo-translate slash command as shortcut for translation skill
- Add /roo-resolve-conflicts slash command as shortcut for conflict resolution skill
The existing translate and merge-resolver modes are preserved. These new skills
and commands provide reusable access to the same functionality.
Closes CLO-722
* feat: add guidances directory with translator guidance file
- Add .roo/guidances/roo-translator.md for brand voice, tone, and word choice guidance
- Update roo-translation skill to reference the guidance file
The guidance file serves as a placeholder for translation style guidelines
that will be interpolated at runtime.
* fix: rename guidances directory to guidance (singular)
* fix: remove language-specific section from translator guidance
The guidance file should focus on brand voice, tone, and word choice only.
* fix: remove language-specific guidelines section from skill file
* Update .roo/skills/roo-translation/SKILL.md
Co-authored-by: roomote[bot] <219738659+roomote[bot]@users.noreply.github.com>
---------
Co-authored-by: Roo Code <roomote@roocode.com>
Co-authored-by: Bruno Bergher <bruno@roocode.com>
Co-authored-by: roomote[bot] <219738659+roomote[bot]@users.noreply.github.com>
* feat: add Claude Opus 4.6 support across all providers (#11224)
* feat: add Claude Opus 4.6 support across all providers
Add Claude Opus 4.6 (claude-opus-4-6) model definitions and 1M context
support across Anthropic, Bedrock, Vertex AI, OpenRouter, and Vercel AI
Gateway providers.
- Anthropic: 128K max output, /5 pricing, 1M context tiers
- Bedrock: anthropic.claude-opus-4-6-v1:0 with 1M context + global inference
- Vertex: claude-opus-4-6 with 1M context tiers
- OpenRouter: prompt caching + reasoning budget sets
- Vercel AI Gateway: Opus 4.5 and 4.6 added to capability sets
- UI: 1M context checkbox for Opus 4.6 on all providers
- i18n: Updated 1M context descriptions across 18 locales
Also adds Opus 4.5 to Vercel AI Gateway (previously missing) and
OpenRouter maxTokens overrides for Opus 4.5/4.6.
Closes#11223
* fix: apply tier pricing when 1M context is enabled on Bedrock
When awsBedrock1MContext is enabled for tiered models like Opus 4.6,
also apply the 1M tier pricing (inputPrice, outputPrice, cache prices)
instead of only updating contextWindow. This ensures cost calculations
and UI display use the correct >200K rates.
* feat: add gpt-5.3-codex model to OpenAI Codex provider (#11225)
feat: add gpt-5.3-codex model and make it default for OpenAI Codex provider
Co-authored-by: Roo Code <roomote@roocode.com>
* fix: prevent parent task state loss during orchestrator delegation (#11281)
* fix: make removeClineFromStack() delegation-aware to prevent orphaned parent tasks (#11302)
* fix: make removeClineFromStack() delegation-aware to prevent orphaned parent tasks
When a delegated child task is removed via removeClineFromStack() (e.g., Clear
Task, navigate to history, start new task), the parent task was left orphaned
in "delegated" status with a stale awaitingChildId. This made the parent
unresumable without manual history repair.
This fix captures parentTaskId and childTaskId before abort/dispose, then
repairs the parent metadata (status -> active, clear awaitingChildId) when
the popped task is a delegated child and awaitingChildId matches.
Parent lookup + updateTaskHistory are wrapped in try/catch so failures are
non-fatal (logged but do not block the pop).
Closes#11301
* fix: add skipDelegationRepair opt-out to removeClineFromStack() for nested delegation
---------
Co-authored-by: Roo Code <roomote@roocode.com>
* fix(reliability): prevent webview postMessage crashes and make dispose idempotent (#11313)
* fix(reliability): prevent webview postMessage crashes and make dispose idempotent
Closes: #11311
1. postMessageToWebview() now catches rejections from
webview.postMessage() so that messages sent after the webview is
disposed do not surface as unhandled promise rejections.
2. dispose() is guarded by a _disposed flag so that repeated calls
(e.g. during rapid extension deactivation) are no-ops.
3. CloudService mock in ClineProvider.spec.ts updated to include
off() — a pre-existing gap exposed by the new dispose test.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: add early _disposed check in postMessageToWebview
Skip the postMessage call entirely when the provider is already disposed,
avoiding unnecessary try/catch execution. Added test coverage for this path.
* chore: trigger CI
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: daniel-lxs <ricciodaniel98@gmail.com>
* fix: resolve race condition in new_task delegation that loses parent task history (#11331)
* fix: resolve race condition in new_task delegation that loses parent task history
When delegateParentAndOpenChild creates a child task via createTask(), the
Task constructor fires startTask() as a fire-and-forget async call. The child
immediately begins its task loop and eventually calls saveClineMessages() →
updateTaskHistory(), which reads globalState, modifies it, and writes back.
Meanwhile, delegateParentAndOpenChild persists the parent's delegation
metadata (status: 'delegated', delegatedToId, awaitingChildId, childIds) via
a separate updateTaskHistory() call AFTER createTask() returns.
These two concurrent read-modify-write operations on globalState race: the
last writer wins, overwriting the other's changes. When the child's write
lands last, the parent's delegation fields are lost, making the parent task
unresumable when the child finishes.
Fix: create the child task with startTask: false, persist the parent's
delegation metadata first, then manually call child.start(). This ensures
the parent metadata is safely in globalState before the child begins writing.
* docs: clarify Task.start() only handles new tasks, not history resume
* fix: serialize taskHistory writes and fix delegation status overwrite race (#11335)
Add a promise-chain mutex (withTaskHistoryLock) to serialize all
read-modify-write operations on taskHistory, preventing concurrent
interleaving from silently dropping entries.
Reorder reopenParentFromDelegation to close the child instance
before marking it completed, so the abort path's stale 'active'
status write no longer overwrites the 'completed' state.
Covered by new tests: RPD-04/05/06, UTH-02/04, and a full mutex
concurrency suite.
* Fix task resumption in the API module (#11369)
* chore: clean up repo-facing mode rules (#11410)
* fix: add maxReadFileLine to ExtensionState type for webview compatibility
---------
Co-authored-by: roomote[bot] <219738659+roomote[bot]@users.noreply.github.com>
Co-authored-by: Roo Code <roomote@roocode.com>
Co-authored-by: Daniel <57051444+daniel-lxs@users.noreply.github.com>
Co-authored-by: Matt Rubens <mrubens@users.noreply.github.com>
Co-authored-by: Chris Estreich <cestreich@gmail.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: Bruno Bergher <bruno@roocode.com>
Co-authored-by: 0xMink <dennis@dennismink.com>
Co-authored-by: daniel-lxs <ricciodaniel98@gmail.com>
Revert to pre-AI-SDK state (commit 67e568f6b)
This commit reverts the codebase to the state before AI SDK migration work began.
Target commit: 67e568f6b - refactor: replace fetch_instructions with skill tool and built-in skills (#10913)
Date: January 29, 2026
This removes approximately 152 commits of AI SDK migration work.
A follow-up PR will add back bug fixes and features that are unrelated to AI SDK.
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix: add character limit to prevent terminal output context explosion
- Enhanced truncateOutput function to accept character limits alongside line limits
- Character limits take priority over line limits to prevent context window explosion
- Added terminalOutputCharacterLimit setting (default: 100,000 characters)
- Updated all terminal output processing to use both limits
- Added comprehensive tests for character limit functionality
Fixes#5775
* feat: add terminal output character limit setting to UI
- Add character limit slider to Terminal Settings UI (default: 50,000)
- Update ExtensionStateContext to manage character limit state
- Add validation for positive character limit values
- Add English translation and translations for all 17 supported languages
- Connect UI to backend through proper message handling
- Character limit takes precedence over line limit to prevent memory issues
* fix: update test expectations for character limit edge cases
- Fix multi-byte character test to account for JavaScript's string length behavior
- Fix newline content test to match actual slice behavior
- Tests now correctly validate the truncateOutput function's character limit handling
---------
Co-authored-by: Roo Code <roomote@roocode.com>
Co-authored-by: Daniel Riccio <ricciodaniel98@gmail.com>
- Allow specific binary formats (.pdf, .docx, .ipynb) to be processed by extractTextFromFile
- Block unsupported binary files with existing "Binary file" notice
- Update tests to cover both supported and unsupported binary file scenarios
- Refactor test mocks for better maintainability and coverage
Follow-up to #2562 adding support for backspace character compression.
Optimizes terminal output by handling backspace characters similar to
carriage returns, improving readability of progress spinners and other
terminal output that uses backspace for animation.
- Added processBackspaces function using efficient indexOf approach
- Added comprehensive test suite for backspace handling
- Integrated with terminal output compression pipeline
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
Co-authored-by: Eric Wheeler <roo-code@z.ewheeler.org>
* fix(terminal): Ensure correct handling of carriage returns for progress bars
This commit refines the tests for `TerminalProcess` to ensure the correct interpretation of terminal output containing carriage returns (`\\r`), which is essential for properly handling dynamic elements like progress bars (e.g., `tqdm`).
- Validated the `processCarriageReturns` method's behavior in simulating terminal line overwrites caused by `\\r`.
- Corrected the expectation in the `handles carriage returns in mixed content` test to accurately reflect the method's output (final line content + preserved escape sequences), confirming the logic works as intended for progress-bar-like updates.
- Fixed a minor Jest `toBe` syntax error in a related test case.
- Suppressed an expected `console.warn` in the non-shell-integration test for cleaner logs.
By ensuring `processCarriageReturns` is correctly tested, we increase confidence that the component responsible for pre-processing terminal output handles progress bars appropriately before the output is potentially used elsewhere (e.g., sent to an LLM).
* fix(test): Make TerminalProcess integration test reliable
This commit fixes the flaky test case `integrates with getUnretrievedOutput to handle progress bars` in `TerminalProcess.test.ts`.
The test previously failed intermittently due to:
1. Relying on a fixed `setTimeout` duration to wait for asynchronous stream processing, which created a race condition.
2. Incorrectly assuming that `await terminalProcess.run(...)` would return the final output directly via its resolved value.
The fix addresses these issues by:
- Removing the unreliable intermediate check based on `setTimeout`.
- Modifying the test to correctly obtain the final output by listening for the `completed` event emitted by `TerminalProcess`, which is the intended way to receive the result.
This ensures the test accurately reflects the behavior of `TerminalProcess` and is no longer prone to timing-related failures.
* Add changeset for terminal carriage return fix
* Implement terminal compress progress bar feature
This commit introduces a new feature to compress terminal output by processing carriage returns. The `processCarriageReturns` function has been integrated into the `Terminal` class to handle progress bar updates effectively, ensuring only the final state is displayed.
Additionally, the `terminalCompressProgressBar` setting has been added to the global settings schema, allowing users to enable or disable this feature.
Tests have been updated to validate the new functionality and ensure correct behavior in various scenarios.
A Benchmark is also added to test the performance.
Not that there is still no i18n support for this.
* Add i18n support for compressProgressBar setting in multiple languages
* Optimize processCarriageReturns function for performance and multi-byte character handling
This commit enhances the `processCarriageReturns` function by implementing in-place string operations to improve performance, especially with large outputs. Key features include:
- Line-by-line processing to maximize chunk handling.
- Use of string indexes and substring operations instead of arrays.
- Single-pass traversal of input for efficiency.
- Special handling for multi-byte characters to prevent corruption during overwrites.
Additionally, tests have been updated to validate the new functionality, ensuring correct behavior with various character sets, including emojis and non-ASCII text.
Highly Density CR case is added to Benchmark
* slight performance improvement by caching several variable
* Optimize multi-byte character handling in processCarriageReturns
Refactor the logic within the `processCarriageReturns` function to simplify the detection of partially overwritten multi-byte characters (e.g., emojis).
Removed redundant checks and clarified the conditions for identifying potential character corruption during carriage return processing. This improves code readability and maintainability while preserving the original functionality of replacing potentially corrupted characters with a space.
Also enforced consistent use of semicolons for improved code style.
* docs: standardize carriage return (\r) and line feed (\n) terminology
Improve code clarity by consistently adding escape sequence notation to all
references of carriage returns and line feeds throughout documentation and tests.
This makes the code more readable and avoids ambiguity when discussing these special characters.
* feat: Improve terminal output processing clarity and settings UI
- Add detailed comments to `processCarriageReturns` explaining line feed handling.
- Relocate `terminalCompressProgressBar` setting below `terminalOutputLineLimit` for better context in UI.
* Fix: Compress Progress Bar Setting Checkbox
---------
Co-authored-by: Matt Rubens <mrubens@users.noreply.github.com>
* fix: addLineNumbers handling of empty content
Empty files should not have line numbers, but non-empty files with empty content at a specific line offset should.
- If content is empty, return empty string for empty files
- If content is empty but startLine > 1, return line number for empty content at that offset
This ensures that the model does not think the file contains a single empty line.
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
* refactor: improve readFileTool XML output format
- Remove unnecessary XML indentation that could confuse the model
- Separate file content from notices and errors using dedicated tags
- Add line range information to content tags
- Handle empty files properly with self-closing tags
- Add comprehensive test coverage
Fixes#2278
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
* fix: always show line numbers in read_file XML output
- Always display line numbers in non-range reads
- Improve XML formatting with consistent newlines for better readability
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
* test: update tests to match new XML format with line numbers
- Update test expectations to match the new XML format with newlines
- Update tests to expect line numbers attribute in content tags
- Modify test assertions to check for the correct line range values
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
* fix: consistent blank line handling in addLineNumbers
- Add newline to all output
- Handle trailing newlines and empty lines consistently
- Add test cases for blank lines:
- Multiple blank lines within content
- Multiple trailing blank lines
- Only blank lines with offset
- Trailing newlines
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
* test: use actual addLineNumbers in read-file-xml tests
- Modified extract-text mock to preserve actual addLineNumbers implementation
- Removed mock implementation of addLineNumbers
- Updated test data to account for trailing newline
- Removed unnecessary mock verification
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
* test: ensure actual addLineNumbers function is called in tests
- Replace direct mocking of addLineNumbers with spy on actual implementation
- Add verification to ensure the real function is called when appropriate
- Add skipAddLineNumbersCheck option for cases where function should not be called
- Update test cases to use appropriate verification options
- Fix numberedFileContent to include trailing newline for consistency
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
* fix: modify readLines to process data directly instead of line by line
- Direct data processing provides more accurate results by preserving exact content with carriage returns
- Improved performance through minimal buffering and efficient string operations
- Use string indexes to find newlines while maintaining their original format
- Handle all edge cases correctly with preserved line endings
- Add tests for various edge cases including empty files, single lines, and different line endings
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
* test: remove unused mockInputContent variable
Remove unused variable declaration to appease ellipsis-dev linter requirements.
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
---------
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
Co-authored-by: Eric Wheeler <roo-code@z.ewheeler.org>
Implement applyRunLengthEncoding function to compress repeated lines in text output:
- Add line repetition compression with count message
- Focus on single line repetitions
- Only compress when beneficial
- Add tests for empty input and single line repetitions
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
Use string indices to find line boundaries instead of splitting into array.
This avoids creating large arrays in memory when truncating big inputs.
- Replace split/join with indexOf/lastIndexOf for line counting
- Use slice to extract start/end sections directly from string
- Maintain same 20/80 ratio for before/after content
Signed-off-by: Eric Wheeler <roo-code@z.ewheeler.org>
This reverts commit 7eee3e0878.
Middle-out truncation is a really great feature and it should still be
implemented, however it unnecessarily interferes with #1365 because it
hooked into the low-level chunk management that comes directly from VSCE
shell integration.
The best place to hook OutputBuilder is as follows depending on the
state of terminal interaction:
1. Foreground terminals:
Cline.ts:
executeCommandTool(...) {
process.on("line", (line) => {
lines.push(line)
...
}
}
2. For background terminals: hook in at the point that getUnretrievedOutput is consumed for active or
inactive terminals in Cline.ts:getEnvironmentDetails()
Please note:
The Terminal classes are very sensitive to change, partially because of
the complicated way that shell integration works with VSCE, and
partially because of the way that Cline interacts with the Terminal*
class abstractions that make VSCE shell integration easier to work with.
At the point that PR#1365 is merged, it is unlikely that any Terminal*
classes will need to be modified substantially. Generally speaking, we
should think of this is a stable interface and minimize changes.
Reverts: #1390
