* fix(tui): expand paste markers when opening external editor
Add getExpandedText() method to Editor that substitutes paste markers
with actual content. Use it in Ctrl-G external editor flow so users
see full pasted content instead of [paste #N ...] placeholders.
* docs: add changelog entries for #444
Add new slash commands for gracefully exiting the interactive mode:
- /quit - exit the application
- /exit - alias for /quit (common CLI convention)
Both commands await shutdown() which emits session_shutdown events
to hooks and custom tools before exiting. Unlike Ctrl+C (twice) which
uses void (fire-and-forget), these commands properly wait for hooks
and tools to complete their shutdown handlers.
JPEG/GIF/WebP images were not rendering in terminals using the Kitty
graphics protocol (Kitty, Ghostty, WezTerm) because it requires PNG
format (f=100). Non-PNG images are now converted to PNG using sharp
before being sent to the terminal.
- Hook commands now execute immediately during streaming (they manage their own LLM interaction via pi.sendMessage())
- File-based slash commands are expanded and queued via steer/followUp during streaming
- prompt() accepts new streamingBehavior option ('steer' or 'followUp') for explicit queueing during streaming
- steer() and followUp() now expand file-based slash commands and error on hook commands
- RPC prompt command accepts optional streamingBehavior field
- Updated docs: rpc.md, sdk.md, CHANGELOG.md
fixes#420
The async diff preview computation could race with tool execution,
causing editDiffPreview to contain an error (file already modified)
while the actual tool result had the correct diff in details.diff.
Fix: prioritize result.details.diff over editDiffPreview when the
tool has executed successfully. The preview is only used before
tool execution completes.
Use !!command to execute bash commands that are shown in the TUI and
saved to session history but excluded from LLM context, compaction
summaries, and branch summaries.
- Add excludeFromContext field to BashExecutionMessage
- Filter excluded messages in convertToLlm()
- Parse !! prefix in interactive mode
- Use dim border color for excluded commands
fixes#414
Claude requires max_tokens > thinking.budget_tokens. When caller specifies
a small maxTokens (e.g. compaction with ~13k tokens) and reasoning is enabled
with high budget (16k tokens), the constraint was violated.
Fix: In mapOptionsForApi, add thinkingBudget on top of caller's maxTokens
(capped at model.maxTokens). If still not enough room, reduce thinkingBudget
to leave space for output.
Applied to both anthropic-messages and google-gemini-cli APIs.
Also adds test utilities for OAuth credential resolution and tests for
compaction with thinking models.
fixes#413
