Cherry-pick from d6acf1a with manual merge resolution.
Keep onboarding edits in draft state until users choose Done or Save and
Exit, so backing out or discarding the wizard no longer persists partial
changes.
Co-Authored-By: Jason Zhao <144443939+JasonZhaoWW@users.noreply.github.com>
--workspace and --config now work as initial defaults in interactive mode:
- The wizard starts with these values pre-filled
- Users can view and modify them in the wizard
- Final saved config reflects user's choices
This makes the CLI args more useful for interactive sessions while
still allowing full customization through the wizard.
- Add `_SELECT_FIELD_HINTS` for select fields with predefined choices
(e.g., reasoning_effort: low/medium/high with hint text)
- Add `_select_with_back()` using prompt_toolkit for custom key bindings
- Support Escape and Left arrow keys to go back in menus
- Apply to field config, provider selection, and channel selection menus
- Add model_info.py module with litellm-based model lookup
- Provide autocomplete suggestions for model names
- Auto-fill context_window_tokens when model changes (only at default)
- Add "Get recommended value" option for manual context lookup
- Dynamically load provider keywords from registry (no hardcoding)
Resolves#2018
Adjust the spawn tool description to keep the workspace-organizing hint while
avoiding language that sounds like the system automatically assigns a dedicated
working directory for subagents.
Made-with: Cursor
MCP servers (e.g. Zapier) return JSON Schema union types like
`"type": ["string", "null"]` for nullable parameters. The existing
`validate_params()` and `cast_params()` methods expected only simple
strings as `type`, causing `TypeError: unhashable type: 'list'` on
every MCP tool call with nullable parameters.
Add `_resolve_type()` helper that extracts the first non-null type
from union types, and use it in `_cast_value()` and `_validate()`.
Also handle `None` values correctly when the schema declares a
nullable type.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The root cause of "Pool timeout" errors is that long-polling (getUpdates)
and outbound API calls (send_message, send_photo, etc.) shared the same
HTTPXRequest pool — polling holds connections indefinitely, starving sends
under concurrent load (e.g. cron jobs + user chat).
- Split into two independent pools: API calls (default 32) and polling (4)
- Expose connection_pool_size / pool_timeout in TelegramConfig for tuning
- Add _call_with_retry() with exponential backoff (3 attempts) on TimedOut
- Apply retry to _send_text and remote media URL sends
Addresses review feedback: moves schedule formatting and state
formatting into dedicated static methods, removes duplicate
in-loop imports, and simplifies _list_jobs() to a clean loop.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
_list_jobs() only displayed job name, id, and schedule kind (e.g. "cron"),
omitting the actual timing and run state. The agent couldn't answer
"when does this run?" or "did it run?" even though CronSchedule and
CronJobState had all the data.
Now surfaces:
- Cron expression + timezone for cron jobs
- Human-readable interval for every jobs
- ISO timestamp for one-shot at jobs
- Enabled/disabled status
- Last run time + status (ok/error/skipped) + error message
- Next scheduled run time
Fixes#1496
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix double bold markers (****) when heading text already contains **
- Strip markdown formatting (**bold**, *italic*, ~~strike~~) from table cells
since Feishu table elements do not support markdown rendering
Fixes rendering issues where:
1. Headings like '**text**' were rendered as '****text****'
2. Table cells with '**bold**' showed raw markdown instead of plain text
Remove the in-progress reaction (reactEmoji) and optionally add a
done reaction (doneEmoji) when the final response is sent, so users
get visual feedback that processing has finished.
Signed-off-by: Sihyeon Jang <sihyeon.jang@navercorp.com>
Replace manual _active_spinner + _pause_spinner/_resume_spinner with
_ThinkingSpinner class that owns the spinner lifecycle via __enter__/
__exit__ and provides a pause() context manager for temporarily
stopping the spinner during progress output.
Benefits:
- Restores Pythonic context manager pattern matching original code
- Eliminates duplicated start/stop boilerplate between single-message
and interactive modes
- pause() context manager guarantees resume even if print raises
- _active flag prevents post-teardown resume from async callbacks
The Rich console.status() spinner ('nanobot is thinking...') was not
cleared when tool call progress lines were printed during processing,
causing overlapping/garbled terminal output.
Replace the context-manager approach with explicit start/stop lifecycle:
- _pause_spinner() stops the spinner before any progress line is printed
- _resume_spinner() restarts it after printing
- Applied to both single-message mode (_cli_progress) and interactive
mode (_consume_outbound)
Closes#1956
Replace fire-and-forget consolidation with archive_messages(), which
retries until the raw-dump fallback triggers — making it effectively
infallible. /new now clears the session immediately and archives in
the background. Pending archive tasks are drained on shutdown via
close_mcp() so no data is lost on process exit.
PR #881 (commit 755e424) fixed the race condition between normal consolidation
and /new consolidation, but did so by making /new wait for consolidation to
complete before returning. This hurts user experience - /new should be instant.
This PR restores the original immediate-return behavior while keeping safety:
1. **Immediate return**: Session clears and user sees "New session started" right away
2. **Background archival**: Consolidation runs in background via asyncio.create_task
3. **Serialized consolidation**: Uses the same lock as normal consolidation via
`memory_consolidator.get_lock()` to prevent concurrent writes
If consolidation fails after session clear, archived messages may be lost.
This is acceptable because:
- User already sees the new session and can continue working
- Failure is logged for debugging
- The alternative (blocking /new on every call) hurts UX for all users
Phase 1 _decide() now includes "Current date/time: YYYY-MM-DD HH:MM UTC"
in the user prompt and instructs the LLM to use it for time-aware scheduling.
Without this, the LLM defaults to 'run' for any task description regardless
of whether it is actually due, defeating Phase 1's pre-screening purpose.
Closes#1929
Instead of adding a separate load_skill tool to bypass workspace restrictions,
extend ReadFileTool with extra_allowed_dirs so it can read builtin skill paths
while keeping write/edit tools locked to the workspace. Fixes the original issue
for both main agent and subagents.
Made-with: Cursor
When restrictToWorkspace is enabled, the agent cannot read builtin skill
files via read_file since they live outside the workspace. This adds a
dedicated load_skill tool that reads skills by name through the SkillsLoader,
which accesses files directly via Python without the workspace restriction.
- Add LoadSkillTool to filesystem tools
- Register it in the agent loop
- Update system prompt to instruct agent to use load_skill instead of read_file
- Remove raw filesystem paths from skills summary
chengyongruandXubin Ren
Xubin RenandXubin Ren
JilunSun7274andXubin Ren
mamamiyearandXubin Ren
h4nz4andXubin Ren
Javis486andXubin Ren
zhangxiaoyu.yorkandXubin Ren
weipeng0098andXubin Ren
Your NameandXubin Ren
Sihyeon JangandXubin Ren
Jiajun XieandXubin Ren
Peter van EijkandXubin Ren
who96andXubin Ren
riseandXubin Ren
Meng YuhangandXubin Ren
BenandXubin Ren