Add "Solutions" category to consolidate prompt so trial-and-error
workflows that reach a working approach are captured in history for
Dream to persist. Remove overly broad "debug steps" skip rule that
discarded these valuable findings.
When a tool call triggers a mid-turn _stream_end(resuming=True), Feishu
channel unconditionally popped the buffer and closed the streaming card.
The next segment then created a new card, causing duplicate/fragmented
replies.
Now send_delta checks _resuming: if True, it flushes current text to
the existing card but keeps the buffer alive so subsequent segments
append to the same card. Only the final _stream_end (resuming=False)
pops the buffer and closes streaming mode.
Made-with: Cursor
Add ack_message config field to QQConfig (default: Processing...). When non-empty, sends an instant text reply before agent processing begins, filling the silence gap for users. Uses existing _send_text_only method; failure is logged but never blocks normal message handling.
Made-with: Cursor
Fixes#2591
The "nanobot is thinking..." spinner was printing ANSI escape codes
literally in some terminals, causing garbled output like:
?[2K?[32m⠧?[0m ?[2mnanobot is thinking...?[0m
Root causes:
1. Console created without force_terminal=True, so Rich couldn't
reliably detect terminal capabilities
2. Spinner continued running during user input prompt, conflicting
with prompt_toolkit
Changes:
- Set force_terminal=True in _make_console() for proper ANSI handling
- Add stop_for_input() method to StreamRenderer
- Call stop_for_input() before reading user input in interactive mode
- Add tests for the new functionality
Replace single-stage MemoryConsolidator with a two-stage architecture:
- Consolidator: lightweight token-budget triggered summarization,
appends to HISTORY.md with cursor-based tracking
- Dream: cron-scheduled two-phase processor that analyzes HISTORY.md
and updates SOUL.md, USER.md, MEMORY.md via AgentRunner with
edit_file tools for surgical, fault-tolerant updates
New files: MemoryStore (pure file I/O), Dream class, DreamConfig,
/dream and /dream-log commands. 89 tests covering all components.
Introduce a CompositeHook that fans out lifecycle callbacks to an
ordered list of AgentHook instances with per-hook error isolation.
Extract the nested _LoopHook and _SubagentHook to module scope as
public LoopHook / SubagentHook so downstream users can subclass or
compose them. Add `hooks` parameter to AgentLoop.__init__ for
registering custom hooks at construction time.
Closes#2603
Make the fixed-session API surface explicit, document its usage, exclude api/ from core agent line counts, and remove implicit aiohttp pytest fixture dependencies from API tests.
Require a single user message, reject mismatched models, document the OpenAI-compatible API, and exclude api/ from core agent line counts so the interface matches nanobot's minimal fixed-session runtime.
Reject mismatched models and require a single user message so the OpenAI-compatible endpoint reflects the fixed-session nanobot runtime without extra compatibility noise.
* feat(channel): add iMessage integration
Add iMessage as a built-in channel via Photon's iMessage platform.
Local mode (macOS) reads ~/Library/Messages/chat.db via sqlite3 with
attachment support, sends via AppleScript, and handles voice
transcription. Remote mode connects to a Photon endpoint over pure
HTTP with proxy support, implementing send, attachments, tapback
reactions, typing indicators, mark-as-read, polls, groups, contacts,
and health checks.
Paragraph splitting sends each \n\n-separated block as a separate
iMessage bubble. Startup seeds existing message IDs to avoid
replaying old messages on restart.
Tested end-to-end on both modes.
* fix(imessage): address review feedback
- Add retry with exponential backoff for SQLite lock contention
- Escape newlines/tabs in AppleScript to prevent parse failures
- Lower _MAX_MESSAGE_LEN to 6000 (paragraph splitting handles UX)
- Add privacy note in README for Photon remote mode
- Merge poll interval constants; make local mode configurable too
* feat(discord): channel-side read receipt and subagent indicator
- Add 👀 reaction on message receipt, removed after bot reply
- Add 🔧 reaction on first progress message, removed on final reply
- Both managed purely in discord.py channel layer, no subagent.py changes
- Config: read_receipt_emoji, subagent_emoji with sensible defaults
Addresses maintainer feedback on HKUDS/nanobot#2330
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(discord): add both reactions on inbound, not on progress
_progress flag is for streaming chunks, not subagent lifecycle.
Add 👀 + 🔧 immediately on message receipt, clear both on final reply.
* fix: remove stale _subagent_active reference in _clear_reactions
* fix(discord): clean up reactions on message handling failure
Previously, if _handle_message raised an exception, pending reactions
(read receipt + subagent indicator) would remain on the user's message
indefinitely since send() — which handles normal cleanup — would never
be called.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* refactor(discord): replace subagent_emoji with delayed working indicator
- Rename subagent_emoji → working_emoji (honest naming: not tied to
subagent lifecycle)
- Add working_emoji_delay (default 2s) — cosmetic delay so 🔧 appears
after 👀, cancelled if bot replies before delay fires
- Clean up: cancel pending task + remove both reactions on reply/error
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
The on_stream and on_stream_end closures in _dispatch hardcoded
their metadata dicts, dropping channel-specific fields like
message_thread_id. Copy msg.metadata first, then add internal
streaming flags, matching the pattern already used by _bus_progress.
Read serve host, port, and timeout from config by default, keep CLI flags higher priority, and bind the API to localhost by default for safer local usage.
Expose OpenAI-compatible chat completions and models endpoints through a single persistent API session, keeping the integration simple without adding multi-session isolation yet.
* feat(feishu): add streaming support via CardKit PATCH API
Implement send_delta() for Feishu channel using interactive card
progressive editing:
- First delta creates a card with markdown content and typing cursor
- Subsequent deltas throttled at 0.5s to respect 5 QPS PATCH limit
- stream_end finalizes with full formatted card (tables, rich markdown)
Also refactors _send_message_sync to return message_id (str | None)
and adds _patch_card_sync for card updates.
Includes 17 new unit tests covering streaming lifecycle, config,
card building, and edge cases.
Made-with: Cursor
* feat(feishu): close CardKit streaming_mode on stream end
Call cardkit card.settings after final content update so chat preview
leaves default [生成中...] summary (Feishu streaming docs).
Made-with: Cursor
* style: polish Feishu streaming (PEP8 spacing, drop unused test imports)
Made-with: Cursor
* docs(feishu): document cardkit:card:write for streaming
- README: permissions, upgrade note for existing apps, streaming toggle
- CHANNEL_PLUGIN_GUIDE: Feishu CardKit scope and when to disable streaming
Made-with: Cursor
* docs: address PR 2382 review (test path, plugin guide, README, English docstrings)
- Move Feishu streaming tests to tests/channels/
- Remove Feishu CardKit scope from CHANNEL_PLUGIN_GUIDE (plugin-dev doc only)
- README Feishu permissions: consistent English
- feishu.py: replace Chinese in streaming docstrings/comments
Made-with: Cursor
When LLM generates faster than channel can process, asyncio.Queue
accumulates multiple _stream_delta messages. Each delta triggers a
separate API call (~700ms each), causing visible delay after LLM
finishes.
Solution: In _dispatch_outbound, drain all queued deltas for the same
(channel, chat_id) before sending, combining them into a single API
call. Non-matching messages are preserved in a pending buffer for
subsequent processing.
This reduces N API calls to 1 when queue has N accumulated deltas.
