Add the 'Crosscheck Work' experiment to the Agents settings. When enabled,
a reviewer sub-agent is spawned after the main agent finishes a turn (with
no pending tool calls). The reviewer critiques the output using a dedicated
system prompt focused on correctness, simplicity, and code quality. If the
reviewer does not respond with 'LGTM!', its feedback is injected as a
synthetic user query back to the main agent, which must address it. This
loop continues until the reviewer approves or max iterations is reached.
Components:
- Feature flag: CrosscheckWork (enabled in DOGFOOD_FLAGS)
- Settings: agents.experiments.crosscheck_enabled,
agents.experiments.crosscheck_model_id,
agents.experiments.crosscheck_max_iterations
- Settings UI: new 'Experiments' subpage under Agents
- Crosscheck module: app/src/ai/crosscheck/ with prompt, reviewer model
- Controller integration: hooks into AfterStreamFinished when no actions
are queued, triggers reviewer, handles feedback injection
- Provider support: OpenAI-compatible and Bedrock direct invocation
- Safety: max iteration guard, error handling, reset on new user query
- Update version from 1.6.3 to 2.0.0 in app/Cargo.toml and Cargo.lock
- Add install-galaxy.sh upload step to build-and-deploy-hermes script
- Include pending AI provider and agent changes
- Construct ServerTask directly for Bedrock CLI subagents so messages route
correctly without needing a server CreateTask upgrade
- Handle CliAgentUserQuery input type in both request translators, including
running command context and terminal output
- Add force_cancel_all_streaming_exchanges fallback for when Ctrl+C finds no
in-flight streams (stuck subagent / unexpected stream end)
- Cancel active conversation on Ctrl+C in agent view compose state
- Skip agent view entry when agent is tagged-in for a running command
- Set root_task_id on Bedrock requests for proper optimistic task upgrade
- Persist app state on will_terminate to avoid losing sessions
- Trust persisted CWD without is_dir() recheck (fixes network mount restore)
- Log warning instead of silently dropping tabs with unreadable root nodes
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix start_agent tool mapping: route to Tool::StartAgent instead of
dead-end Tool::Subagent so tool calls become executable actions
- Block start_agent until child finishes: parent waits for child
conversation to complete and receives full output as tool result
- Fix tool result delivery: add StartAgent/StartAgentV2 cases to
extract_tool_result_content so the model actually sees agent output
- Support parallel agent spawning: change StartAgent action phase from
Serial to Parallel, and track multiple pending agents via Vec
- Mark child conversations as Success on EndTurn: emit
ConversationStatus::Success when a child stream ends with no actions
- Skip orchestration SSE in local mode: prevent app freeze from trying
to connect to non-existent server
- Remove send_message_to_agent and suggest_next_prompt from tool list:
these require server infrastructure that doesn't exist in OSS mode
- Add "Waiting for sub-agents..." status message while agents process
- Fix child agent pane close: actually dismiss instead of re-hiding,
track dismissed IDs to prevent re-creation on restart
- Fix cache_miss_tokens calculation and show per-block cache stats
- Add [tool-debug] logging throughout tool invocation pipeline
Bump version to 1.6.0.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New features:
- Inline subagent panels with expand/collapse and click-to-toggle
- /context slash command to inspect bedrock_message_history
- Child-to-parent question routing with auto-answer for subagents
- Subagent token usage and cost merging into parent conversation
- Randomized session-colored user avatar silhouettes
Bug fixes:
- Bedrock: remove orphaned tool_results after compaction
- Bedrock: self-healing exchange lookup for out-of-order streaming
- Bedrock: append continuation prompt when conversation ends with assistant
- Duration sanity check rejects epoch-time artifacts from session restore
- Cache hit rate calculation uses actual total_input_tokens
- Hide "Time to first token" when value is zero
Improvements:
- Demote verbose bedrock-debug logs to debug/trace levels
- Bedrock tool usage counting falls back to action counting
- Remove logout menu item from workspace menu
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Major features:
- Auto-compact: triggers conversation summarization when context window >= 85%,
compacts Bedrock message history to a summary pair, and tracks live context tokens
- Bedrock summarization: plumbs `is_summarization` flag through translator/client/response
pipeline, handles SummarizeConversation input type, and marks `summarized` in metadata
- Session restore: rebuilds bedrock_message_history from persisted task messages via
newly-public `convert_proto_message`, preventing empty history on reconnect
- Subagent orchestration: adds SubagentQuestion/Answer/CompletionSummary event types,
parent-child question routing with depth limits, retry counting, and drain methods
- Summarization UI: inline SummarizationView in AI blocks with progress/finished states
Refactors:
- Rename WarpTheme → GalaxyTheme across ~100 files (rebrand continuation)
- Rename warp_home_config_dir → galaxy_home_config_dir and related path functions
- Predefined rules: replace "System Defined Rule #N" with descriptive names
(e.g. "Correctness Over Speed", "Never Guess") and add lookup helpers
- Usage view: replace cumulative input/output token display with live context tokens,
cache hit rate calculation, and separate cache read/write stats
- Telemetry: remove verbose doc comments, simplify trait definitions
- Facts view: simplify delete permission check (always allow local deletion)
- Remove warp_managed_paths_watcher.rs (dead code)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Description
Running `/pr-comments` (or any other slash command that targets the currently-selected conversation) while an AI response stream is in-flight crashed in debug builds. `BlocklistAIController::send_request_input` hits its in-flight invariant and fires `safe_assert!(false, ...)` (panics in debug, returns `Err` in release).
The user-query path (`send_query`) avoids this because it pre-cancels any active stream on the target conversation with `CancellationReason::FollowUpSubmitted` before calling `send_request_input`. Slash commands bypass that path and dispatch directly via `SlashCommandRequest::send_request`, so the cancel never happens.
This PR makes `send_slash_command_request` mirror that cancel-and-resend: if the target conversation has an in-flight stream, cancel it before dispatching. All `SlashCommandRequest` variants are conceptually a fresh user turn (a follow-up), so this matches the semantics users already get from typing a follow-up message.
`send_queued_slash_command_request` is unchanged — it's only invoked from `Input::submit_queued_prompt` once the conversation is idle, so no pre-cancel is needed.
## Testing
Verified locally. Demo [here](https://www.loom.com/share/a9638ac9a53b48349ce21d03eaba516a)!
- Manually reproduced the crash on a dogfood debug build by triggering `/pr-comments` mid-stream; confirmed the panic at the `safe_assert!` in `BlocklistAIController::send_request_input`.
- After the fix: same repro cancels the in-flight turn and dispatches `/pr-comments` cleanly. Same behavior verified for `/skills` (InvokeSkill), `/compact` (Summarize), and `/create-environment`.
## Agent Mode
- [x] Warp Agent Mode - This PR was created via Warp's AI Agent Mode
## Changelog Entries for Stable
CHANGELOG-BUG-FIX: Fixed an issue where slash commands sent while an agent was still responding were silently dropped. Now, slash commands like `/pr-comments` run as follow-ups, just like typed messages.
---
Run: https://staging.warp.dev/conversation/aeecfa2d-1456-440b-9961-c27295d82531
Plan: https://staging.warp.dev/drive/notebook/Ryj2w6xDSqDwNJYhpwwI0Vly
rkw6086
Ryan Ward