* Add the FeatureFlagService to the SDK
* Fix comments
* Dispose of the telemetry service
* Dispose of the feature flag service
* Address PR feedback
* Stop the polling early if a new one is triggered with another user id
* Address PR feedback
* Dispose of the feature flag service
---------
Co-authored-by: Saoud Rizwan <7799382+saoudrizwan@users.noreply.github.com>
* test(sdk): add regression tests for structured ToolOperationResult truncation
MessageBuilder tests only covered string and {type:
🔮 View transcript: https://agentlogs.beatrixwoo.workers.dev/s/mrbt2b39jfn370scr4g0ivzo"text"} tool-result
content, not the structured ToolOperationResult[] shape the default tools
(run_commands, read_files, search_codebase) actually emit. Those entries
are plain {query, result, success} objects with no type discriminator, so
the token-bloat path they create was unprotected by tests.
Adds regression tests using the real structured shape: huge result, huge
query, huge read_files payload, aggregate budget across multiple results,
mutation safety, and provider-formatted AI SDK payload size. Assertions
are on actual serialized payload sizes, not transcript shape.
The new tests fail at this commit by design; the following commit makes
them pass.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(sdk): truncate structured ToolOperationResult strings in MessageBuilder
The runtime stores structured tool outputs (ToolOperationResult[] from
run_commands/read_files/search_codebase) directly as the tool_result
content array (agentPartToContentBlock casts the array straight through).
Those entries have no type discriminator, so MessageBuilder's per-result
truncation, aggregate byte counting, and budget truncation all skipped
them — multi-megabyte command outputs and file reads were JSON-serialized
in full into every subsequent provider request.
MessageBuilder now deep-truncates nested strings inside structured
entries (middle truncation, preserving head and tail), counts them
against the aggregate text budget, collects them as budget-truncation
candidates, and deep-clones them before mutation so the original
conversation history stays untouched. Image blocks are skipped so base64
payloads survive intact.
Real-inference A/B on openrouter:minimax/minimax-m2.7 with realistic
structured payloads: 58.7% overall input-token reduction (82.6% on a
single huge command output) with identical answer correctness.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
🔮 View transcript: https://agentlogs.beatrixwoo.workers.dev/s/mrbt2b39jfn370scr4g0ivzo
* fix(sdk): include fetch_web_content in MessageBuilder truncation targets
Review feedback: fetch_web_content also returns ToolOperationResult[] and
its executor allows responses up to 5MB, but the tool was missing from
TARGET_TOOL_NAMES, so a single web fetch could still bloat every
subsequent provider request. Adds the tool to the truncation target set
with a regression test.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
🔮 View transcript: https://agentlogs.beatrixwoo.workers.dev/s/mrbt2b39jfn370scr4g0ivzo
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The run_commands tool result's query field repeated the entire executed
command, which already exists verbatim in the assistant tool-call input.
For large generated-file commands (e.g. cat <<EOF heredocs) this
duplicated thousands of chars of source text into every subsequent
provider request.
Bound the provider-facing echo to a 200-char preview plus a truncation
note pointing at the tool call input. Short commands pass through
unchanged. Applies to both createBashTool and createWindowsShellTool,
on success and error paths.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
🔮 View transcript: https://agentlogs.beatrixwoo.workers.dev/s/og840mbjqaog4zita8m0262m
The SDK 0.50.1 upgrade widened convertToOpenAiMessages to take
Anthropic.Messages.MessageParam[], which left the ClineStorageMessage
import referenced only in comments. tsc does not flag unused imports in
this config, but biome lint does, and it blocked the 3.89.2 publish.
VS Code 1.123.0 bumped its bundled runtime from Node 22 to Node 24
(Electron 39 to 42). The Anthropic provider broke on the updated editor
because the old SDK (<=0.41.x) shipped a legacy runtime built on
node-fetch and an internal _shims layer that does not work under Node 24.
0.50.1 is the first SDK release rewritten on top of the platform's native
fetch: it has zero runtime dependencies (no node-fetch, no _shims), which
removes the incompatibility. This is the actual fix; the earlier 0.40.1
bump did not change the runtime architecture.
The 0.50.1 type changes are minimal:
- Usage gained a required server_tool_use field, so fabricated Usage
objects in the gemini/o1/openai/vscode-lm transforms set it to null.
- ContentBlockParam widened, so Anthropic.MessageParam is no longer
structurally assignable to ClineStorageMessage. Handled by narrowing
the two transform helpers that only ever receive Cline history
(sanitizeAnthropicMessages, convertAnthropicMessageToGemini), typing
getSavedApiConversationHistory as the Cline history it reads, and
narrowing ContextManager's loosely-typed truncated output back to
ClineStorageMessage at the two provider/hook boundaries.
VS Code 1.123.0 bumped its bundled runtime from Node 22 to Node 24
(Electron 39 to 42). The extension passes VS Code's globalThis.fetch to
every provider SDK, but @anthropic-ai/sdk was pinned at 0.37.0, which
predates the SDK's native-fetch rewrite and relies on legacy _shims
runtime detection that breaks under Node 24. The modern OpenAI and Gemini
SDKs are unaffected, which is why only the Anthropic provider broke after
users updated VS Code.
Bump to ^0.40.1, the first release with the native-fetch rewrite that
restores Node 24 compatibility, while staying short of the latest line's
larger breaking surface.
The only code change the bump requires is narrowing the image source type:
ImageBlockParam.source widened from a base64-only type to
Base64ImageSource | URLImageSource. Add a getBase64ImageSource/
getImageDataUrl helper in shared/messages/content.ts and route the
provider transforms through it. Cline only ever produces base64 image
sources, so behavior is unchanged; the helper emits the same data URL the
inline code did.
* feat: Enforce a production singleton Cline Hub
This PR changes local Hub startup/discovery so production uses one stable daemon per user machine instead of silently creating additional hubs on random ports.
Replace resolveSharedHubOwnerContext with resolveProductionHubOwnerContext
across doctor and hub server lifecycle management to scope hub discovery
to the production owner.
Additionally:
- Preserve and propagate auth tokens when retiring incompatible hubs
- Throw a clear error when a compatible hub is already running but its
discovery record is missing, guiding users to run 'cline doctor fix'
- Gate port fallback behind an explicit allowPortFallback override
- Update tests to mock the new production hub owner context
* patches
* fix
* hasExplicitPort
* Restored daemon cron startup, made discovery auth tokens required again, and fixed graceful hub stop/restart paths to use the selected production/shared owner context.
* clean up
* patches
* fix Polynomial regular expression
* test
* fix: require explicit hub port fallback in production
* fix(cli): stop pgrep from parsing the hub daemon marker as an option
pgrep treats the "--cline-hub-daemon" pattern as an unknown long option
and exits 2, so doctor never found stale daemons from compiled-binary
installs, which are exactly the processes 'cline doctor fix' is told to
clean up. Pass "--" before the pattern to end option parsing.
* fix(hub): retire legacy shared-owner hubs on production startup
Pre-singleton production builds tracked the local hub under the shared
owner discovery path and spawned daemons on random fallback ports. The
production owner context never reads that path, so upgrades would leave
those daemons running indefinitely with no way to reuse or stop them.
Retire the recorded legacy hub (its record carries the auth token and
pid needed for a graceful stop) and clear the legacy record before
resolving the production hub.
* refactor(hub): simplify stale discovery clearing, share capability list
shouldClearStaleHubDiscovery was only ever called with
discoveredVerified=false (the true assignment sits on a return path),
so the expected-hub probe and compatibility check had no effect and the
condition reduced to "a discovery record exists and was not reused".
Replace it with a plain conditional and drop the tests that exercised
unreachable states.
Also move the hub capability list into a typed HUB_CAPABILITIES
constant in @cline/shared next to HubCapabilityName so the server
cannot drift from the type.
---------
Co-authored-by: Saoud Rizwan <7799382+saoudrizwan@users.noreply.github.com>
* fix(cli): suppress flickering console windows on Windows by setting windowsHide on child processes
On Windows, child_process.spawn/execFile default to windowsHide: false,
so console-subsystem children (powershell, rg, git, node, npm) can
allocate a new visible console window - guaranteed when detached: true
is used. In the CLI this caused constant short-lived window flashes
from run_commands, the git status bar polling, ripgrep searches and
indexing, clipboard helpers, and hook/plugin node subprocesses.
Set windowsHide: true (CREATE_NO_WINDOW; a no-op on non-Windows) on all
remaining spawn/spawnSync/execFile call sites in the SDK core, CLI,
Cline Hub, and example plugins, matching the pattern already used by
the MCP client, checkpoint-hooks, and StandaloneTerminalProcess.
* Update apps/cli/src/commands/kanban.ts
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
---------
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* fix(llms): avoid disabled reasoning for fable 5
* fix(llms): route fable reasoning by family
* Revert "fix(llms): route fable reasoning by family"
This reverts commit 6dd4e5dcf5.
* fix(llms): match claude fable reasoning workaround broadly
* fix(core): configured agent support as subagent tools
Introduce configured agent config parsing and tool creation for
subagents. Agent configs are defined via YAML frontmatter files
specifying name, description, tools, skills, model, and system prompt.
- Add `configured-agent-config` for loading and parsing agent
definitions from search paths
- Add configured agent tool factory that wraps delegated agents as
named subagent tools with policy and approval support
* patch
* patches
* fixes
* Infinite loop when YAML block is a non-object fix
* apply feedback
Forwarded host requestToolApproval into configured subagents.
Used the resolved workspace config root for configured-agent skills discovery.
Split configured-agent skill loading from root-session skills enablement.
Added host lifecycle/event plumbing for configured subagents via shared subagent callbacks.
Made UserInstructionConfigService.createSkillsExecutor optional and guarded its use.
* threaded
* test(core): cover configured subagent skill isolation (#11396)
---------
Co-authored-by: Saoud Rizwan <7799382+saoudrizwan@users.noreply.github.com>
The Fable 5 PR (#11385) made claude-fable-5 the newest anthropic model,
which sorts first in the generated catalog. Legacy provider migration
defaults to the first catalog model, so the migrated default changed from
claude-opus-4-8 to claude-fable-5. Update the test expectation to match.
* Improve bug report form: rename surface dropdown, add IDE/CLI diagnostics
Rename the 'Plugin Type' dropdown to 'Cline Surface' since CLI is not a plugin; the option values keep each choice unambiguous.
Add an 'IDE / CLI Diagnostics' field with per-surface copy-paste steps for About info (VSCode Help/About, JetBrains Help/About Copy button) and a CLI exception using 'cline --version'. System Information is left as-is; minor overlap is acceptable.
* Update repo-label-issues workflow for renamed Cline Surface field
The auto-labeler matches the rendered '### Plugin Type' heading. Since the form label was renamed to 'Cline Surface', update the three regexes so JetBrains/VS Code/CLI labels keep applying.
* doc(sdk): add host logger support in plugin examples
Add examples to use the exposed `ctx.logger` to plugins via the `setup` second argument for
diagnostics. Wire logging into the agents-squad example to record setup,
subagent starts, follow-ups, and async failures, with a `logPluginError`
helper that falls back to severity-tagged logs. Update README with
logger usage guidance and examples.
* patches
* fix: empty SDK message content replay for Bedrock CLINE-2373
This fixes SDK message formatting when persisted conversation history contains an empty user or assistant message, such as after an interrupted task is resumed.
Instead of dropping the message turn, the SDK now preserves it and inserts a text content block:
ERROR: EMPTY CONTENT
This prevents providers like Amazon Bedrock from rejecting replayed history with empty content arrays while avoiding message removal that could affect provider turn ordering.
* Exported EMPTY_CONTENT_TEXT from @cline/shared so core/shared use one constant
* fix(cli): connector thread session routing & stale hub session
Fix connector thread session routing and stale hub session recovery
**PR Description**
This fixes connector messages from separate chat threads being routed into the wrong active runtime session.
**Issue**
In Slack, if a user sent a message in a different thread while another thread was still processing, the new message could be treated as a steer message for the active task. Users could also see errors like:
```text
Slack bridge error: session not found: 1780596180501_ms45m
```
when a connector thread had a persisted session id that no longer existed in the hub, such as after a hub restart.
**Cause**
Connector conversation bindings and active turn queues were using participant identity as the primary key in several paths. That allowed messages from the same user in different chat threads to resolve to the same connector session/active turn.
Separately, persisted connector `sessionId` values were trusted without checking whether the hub still had that runtime session. After a hub restart, the connector could try to send input to a stale session id.
**Fix**
- Store connector conversation bindings by thread id instead of participant key.
- Key connector active turn queues by thread id across Slack, Discord, Telegram, Google Chat, Linear, and WhatsApp adapters.
- Only treat a follow-up as a steer message when the active turn belongs to the same thread.
- Keep participant key/label as metadata instead of using it as the conversation binding key.
- Validate a persisted session id with the hub before reusing it.
- If the persisted session is missing, clear it from thread state and start a fresh runtime session.
- Update schedule delivery metadata to target thread ids while preserving participant metadata.
- Add regression coverage for cross-thread active sessions and stale persisted session ids.
**Verification**
```bash
bun -F @cline/cli typecheck
bunx vitest run apps/cli/src/connectors/connector-host.test.ts apps/cli/src/connectors/thread-bindings.test.ts apps/cli/src/connectors/adapters/slack.test.ts apps/cli/src/connectors/adapters/telegram.test.ts apps/cli/src/connectors/adapters/discord.test.ts apps/cli/src/connectors/adapters/gchat.test.ts apps/cli/src/connectors/adapters/linear.test.ts apps/cli/src/connectors/adapters/whatsapp.test.ts
```
* patches
Saoud Rizwan