The extension system currently only forwards agent_start, agent_end,
turn_start, and turn_end events. This means extensions cannot access
streaming text (token-by-token), message lifecycle, or tool execution
progress — all of which are available to internal subscribers.
This adds forwarding for the remaining 6 agent event types:
- message_start, message_update, message_end
- tool_execution_start, tool_execution_update, tool_execution_end
These follow the exact same pattern as the existing forwarded events:
new interfaces in types.ts, exports in index.ts, and else-if blocks
in _emitExtensionEvent(). The new types are included in ExtensionEvent
and automatically flow through RunnerEmitEvent (they're not in the
exclusion list).
This enables extensions to build real-time UIs, streaming WebSocket
bridges, and other integrations that need fine-grained event access.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Mario Zechner <badlogicgames@gmail.com>
After compaction, context token count is unknown until the next LLM
response. Instead of showing stale pre-compaction values or heuristic
estimates, the footer now shows ?/200k.
ContextUsage.tokens and ContextUsage.percent are now number | null
(breaking change). Removed usageTokens, trailingTokens, lastUsageIndex
from ContextUsage (internal details).
Also fixed _checkCompaction() using .find() (first compaction) instead
of getLatestCompactionEntry() (latest), which caused incorrect overflow
detection with multiple compactions.
Closes#1382
Add metadata?: Record<string, unknown> to StreamOptions so providers
can extract fields they understand. Anthropic provider extracts user_id
for abuse tracking and rate limiting. Other providers ignore it.
Based on #1384 by @7Sageer, reworked to use a generic type instead of
Anthropic-specific typing on the base interface.
* extend interleaved thinking test to Anthropic first-party provider
* switch back to global Bedrock model identifier
* set retry to 3 for both
* enable bedrock claude interleaved thinking by default and use completeSimple in test
- Replace verbose ANTIGRAVITY_SYSTEM_INSTRUCTION with compact version from CLIProxyAPI
- Replace bridgePrompt override with [ignore] wrapper pattern
- Switch Antigravity Gemini test model from gemini-3-flash to gemini-3-pro-high
- Rename calculator tool to math_operation (gemini-3-pro ignores schema for 'calculator')
closes#1415
Update extensions.md and rpc.md to accurately reflect that ctx.hasUI is
true in RPC mode. Document missing unsupported/degraded ExtensionUIContext
methods: pasteToEditor, getAllThemes, getTheme, setTheme.
Co-authored-by: Mario Zechner <badlogicgames@gmail.com>
The invalid-params error paths already listed available agents,
but the unknown-agent path in runSingleAgent just returned
'Unknown agent: claude' with no hint what exists.
Now: 'Unknown agent: "claude". Available agents: "worker".'
Model would guess names like "claude", "default", or skill names
like "brave-search". Now it self-corrects on the next call.
Tested with Opus 4.6: without fix, model gave up on subagent
and ran the tool calls itself instead. With fix, 1 wasted call.
Include tool parameter names, types, descriptions in a collapsible
section under each tool in the export HTML. Also adds parameters to
pi.getAllTools() return value.
closes#1407, closes#1416
rg was only downloaded lazily when the grep tool was invoked, but the
LLM often uses rg directly via the bash tool. Ensure both fd and rg
are downloaded at startup so they are available in PATH for all tools.
fixes#1348
- Use parametersJsonSchema instead of parameters for Gemini tool declarations
to support full JSON Schema (anyOf, oneOf, const, etc.)
- Keep legacy parameters field for Claude models on Cloud Code Assist, where
the API translates parameters into Anthropic's input_schema
- Revert claude-opus-4-6-thinking back to claude-opus-4-5-thinking (model
doesn't exist on the Antigravity endpoint)
fixes#1398
Windows does not ship with unzip, causing fd/rg download to fail on
first run in PowerShell. Use tar (bsdtar, available on Windows 10+)
for both .tar.gz and .zip extraction. Also adds proper error checking
on the extraction result.
fixes#1348
Add pasteToEditor(text) method that pastes text into the editor via
bracketed paste sequences, triggering paste handling (including collapse
for large content). Unlike setEditorText which directly replaces content,
pasteToEditor routes through handleInput on the active editor component.
- Add pasteToEditor to ExtensionUIContext interface
- Add handleInput to EditorComponent interface (was missing, all
concrete implementations already had it)
- Implement in interactive mode via bracketed paste sequence
- Add fallback in RPC mode (delegates to setEditorText)
- Document in extensions.md
* fix(coding-agent): update test model from opus-4-5 to opus-4-6
* fix(coding-agent): delegate onEscape/onCtrlD/onPasteImage to defaultEditor
When an extension provides a custom editor via setEditorComponent(),
onEscape, onCtrlD, and onPasteImage were copied by value from
defaultEditor. Later, when retry/compaction/branch-summary temporarily
swapped defaultEditor.onEscape to an abort handler, the custom editor
still held the stale reference, so pressing Escape during retry did
nothing.
Use closures that delegate to defaultEditor at call time, matching the
pattern from f1b1d54 which fixed the same issue for onExtensionShortcut.
* feat(antigravity): update model costs and tokens, migrate compaction tests to Claude 4.6
- Update model costs for various models in `models.generated.ts` to reflect the latest pricing.
- Update maxTokens for certain models.
- Migrate compaction tests from `claude-opus-4-5-thinking` to `claude-opus-4-6-thinking` in `compaction-thinking-model.test.ts`.
* fix: remove unnecessary peer dependencies in package-lock.json
This commit removes the `peer: true` flag from several dependencies in `package-lock.json`. These dependencies (lit, tailwind-merge, tailwindcss, picomatch, @tailwindcss/typescript, vite, picomatch) do not need to be explicitly marked as peer dependencies in this context, as they are already managed as regular dependencies. Removing the `peer: true` flag simplifies the dependency graph and avoids potential conflicts during installation.
setupAutocomplete only sets the provider on defaultEditor. When an
extension customizes the editor via setEditorComponent during
session_start, the custom editor is created before setupAutocomplete
runs, so it never receives the autocomplete provider. This breaks forced
file suggestions, slash commands, and all other completions.
Set the provider on the active editor too when it differs from the
default, covering initial load, reload, and settings changes.
* feat(tui): extract KillRing and UndoStack, add to Input
Extract kill ring and undo logic from Editor into reusable classes:
- KillRing: ring buffer with accumulation for consecutive kills
- UndoStack<S>: generic stack with clone-on-push semantics
Refactor Editor to use both classes. Add kill ring (kill/yank/
yank-pop), undo with coalescing, and deleteWordForward to Input.
* feat(tui): extract handleBackspace() and handleForwardDelete()
Add a manually inserted "auto" model entry for OpenRouter alongside
the existing "openrouter/auto" entry, allowing users to select the
auto-routing model with a shorter identifier.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Claude Opus 4.5 has been replaced by Claude Opus 4.6 on the
Antigravity (Google Cloud Code Assist) platform.
- Update model definition in generate-models.ts
- Update generated models output
