Fix AI loop stalling and large file sync issues

refactor: remove hardcoded ai.jerxie.com references ...

- Dynamically generate CORS origins based on HUB_PUBLIC_URL.\n- Use window.location.origin in frontend provision scripts.\n- Set localhost/127.0.0.1 as default configurations for local nodes, docker-compose, and spawning scripts.

feat(agent): implement dynamic configuration reloading ...

- The Agent Node now actively reloads its config file during the heartbeat retry loop. \n- Supports both original flat YAML and nested 'cortex' token YAML.\n- Agent no longer caches its node_id or auth_token dynamically at startup, drastically improving resilience on edge.

feat(agent): implement resilient auth handshake loop ...

- The Agent Node will no longer crash and exit if the initial token validation fails or the Orchestrator API is unavailable.\n- Replaced sys.exit(1) on SyncConfiguration rejection with a 5-second sleep and retry loop.\n- This allows nodes to transparently recover once a user dynamically updates their background config with the correct token.

fix(mesh): resolve synology shell bug + bandwidth throttle ...

- Changed mesh-terminal-control to fallback to /bin/sh if bash is missing.\n- Added ~25MB/s bandwidth throttle to Hub/Node file sync to prevent gRPC streaming timeouts.\n- Bumped agent version to 1.0.20.

feat(agent): change default FS_ROOT to '/' on Unix to fix NAS access ...

- Default FS_ROOT is now '/' instead of '~'.\n- Resolved traversal block for /volume1 paths on Synology.\n- Bumped agent version to 1.0.19.

feat(agent): support FS_ROOT awareness in chat sessions to fix Nas access ...

- Updated Agent to allow out-of-session file access if path is within FS_ROOT.\n- Bumped agent version to 1.0.18.

feat(agent): implement 'unbrickable' recovery mode ...

- Moves core imports into the retry loop to survive import-time crashes.\n- Ensures the background HTTP updater stays alive even if the main node logic fails.\n- Bumps version to 1.0.17.

feat(provision): dynamic browser support and installer resiliency ...

- Added browser support toggle to one-liner UI.\n- Updated provision endpoint to handle skip_browsers and disable skill in config.\n- Improved installer to handle --skip-browsers and avoid hardcoded URLs.\n- Added safeguards to browser skill to prevent whole-agent crashes.\n- Enabled invite_token auth for agent updates.\n- Bumps version to 1.0.15.

feat(installer): add automatic playwright browser installation ...

- Makes the one-liner truly 'batteries-included' for browser automation.\n- Bumps version to 1.0.13.

fix(installer): look for config in install_dir to prevent identity reset

fix(agent): use Optional type hint for Python 3.8 compatibility ...

- Fixes 'TypeError: unsupported operand type(s) for |' on Synology NAS.\n- Bumps version to 1.0.11 to trigger sync.

fix(agent): resolve config path absolutely and ensure correct CWD for background processes

fix(daemon): support nohup fallback for non-systemd Linux systems ...

- Detects if systemctl --user is functional.\n- Falls back to nohup background loop on NAS/OpenWRT.\n- Adds 'cortex-ctl' script for background management.

fix(bootstrap): add pip bootstrapping and resilient dependency install ...

- Attempt to use 'ensurepip' if pip is missing (common on Synology/OpenWRT).\n- Support --user fallback for dependency installation.\n- Bump version to 1.0.8 to sync improvements to all nodes.

fix(bootstrap): add custom YAML parser to installer for resilience ...

- Allows installer to read agent_config.yaml even if PyYAML is not installed.\n- Fixes 'token required' error on fresh headless nodes (Synology/OpenWRT).

fix(agent): handle read-only mounts in installer and bump to 1.0.7 ...

- Installer now skips files it can't delete (like read-only Docker volumes).\n- This allows test nodes to successfully update.

fix(agent): correct path to bootstrap_installer in updater ...

- Updated the relative path to 3 levels up to correctly find the bootstrapper in the agent-node root.\n- Bumping to 1.0.6.

test: bump version to 1.0.5 to trigger global auto-update test

fix(agent): correctly handle directory structure during extraction ...

- The Hub now wraps the agent bundle in an 'agent-node/' prefix.\n- The installer now only strips the 'agent-node/' prefix if present, preventing it from accidentally deleting the 'src/' folder.

fix(agent): enable INFO logging and bump to v1.0.3 ...

- Unified logging to stdout so update progress is visible in agent.out.log.\n- Triggering update to v1.0.3.

chore: remove test nodes from prod and bump agent version to 1.0.2 ...

- Test nodes are no longer built/ran in the Jerxie production environment.\n- Version bump triggers auto-update on all active agents to pull in the new skills-enabled bundle.

fix(agent): robust daemon install and gRPC config ...

- bootstrap_installer now forces config overwrite during daemon setup to prevent stale token reuse.\n- ai-hub now preserves gRPC port in generated config, preventing agents from defaulting to localhost.\n- installer now ensures gRPC endpoint always has an explicit port.

fix(agent): make bootstrap installer resilient to pip meta errors ...

- Adds --ignore-installed to dependency installation to bypass metadata/RECORD file errors often found in Anaconda and system Python environments on macOS.

fix(agent): separate hub vs node authentication tokens in bootstrap installer ...

- bootstrap_installer now uses 'secret_key' for HTTP hub requests (version/download) and correctly preserves/restores the node-specific 'invite_token' (auth_token) for gRPC.

fix(agent): bootstrap installer auth token precedence ...

- Fix 401 Unauthorized during upgrades by having bootstrap_installer check for `secret_key` in agent_config.yaml before falling back to `auth_token`, since the /api/v1/agent/version endpoint expects the Hub's global SECRET_KEY rather than a node's specific invite token

refactor: organize agent-node client code structure by moving core packages to src/ ...

- Moved agent_node, protos, and shared_core inside src/

- Updated Dockerfile, deployment scripts, compile_protos.sh, installers

- Updated bundle generation in ai-hub to reflect cleaner structure and better README

feat: add daemon/service installation for Mac and Linux ...

- Added install_service.py to configure launchd on macOS and systemd on Linux

- Added --daemon flag to bootstrap_installer.py which acts as a simple 1-click install

feat: add swarm control chat history clear & agent node auto-update system ...

- Added API and UI to clear chat history in Swarm Control while preserving node sync

- Implemented HTTP-based auto-update agent bootstrapper and core daemon

- Removed unused workflow profile and orphaned voice-interaction-handler skill

- Fixed voice chat response duplication and tool-belt instructions

feat(node): add default FS_ROOT pointing to user home

feat: enhance node access control, user preferences, and skill management ...

- Secured node-related endpoints with group-based access control.
- Fixed user preference merging to preserve node attachments.
- Restricted modification and deletion of system skills.
- Hidden administrative node controls for non-admin users.
- Fixed WebSocket connectivity and permission checks for node mesh.
- Filtered node visibility in profile and swarm control to granted nodes only.

fix(agent-node): use different mount point for skills to avoid OCI runtime error

fix(agent-node): remove broken skills symlink causing docker mount failure

Merge branch 'master' of https://gitbucket.jerxie.com/git/yangyangxie/cortex-hub

chore: set restart policy to always for all containers

fix: robust skill discovery for local and docker environments

fix(mesh): sync terminal status codes and implement periodic ghost mirror cleanup

fix(orchestration): resolve skill hallucination, brace crashes, and silent turns in swarm control ...

- Implemented strictly-worded 'Tool-belt' in system prompt to prevent hallucinating external skills.
- Replaced fragile .format() with robust .replace() for system prompts to prevent crashes on curly braces.
- Optimized TTS streaming: increased concurrency and reduced chunk sizes to eliminate mid-sentence pauses.
- Fixed 'silent turn' bug in Swarm Control by ensuring fallback content is yielded even if AI only produces reasoning/tools.
- Improved history formatting to handle assistant turns mapping to tool calls without main content.
- Corrected Gemini model fallback to gemini-1.5-flash.

feat: enhance swarm terminal UI and AI command execution ...

- Implement resizable terminal height in Swarm Control page
- Improve AI command execution with better status tracking and output streaming
- Add documentation for Swarm architecture analysis and AI file sync integration
- Fix terminal scrolling to follow latest output

feat: Truly solve PTY serialization via Automatic Session Multiplexing and Idle Reaper

fix: Solve node thread saturation by serializing PTY session access and fixing cancellation leaks

fix: Solve silent node capacity rejections and improve SubAgent retry logic

feat: Remove DSPy dependency and implement Swarm Control ...

- Removed DSPy-based pipelines (code assistant, file selector, question decider).
- Implemented Swarm Control for multi-node command execution and terminal handling.
- Updated gRPC services for batch command processing and improved perception.
- Replaced Coding Assistant page with Swarm Control interface in the UI.
- Updated Protobuf definitions to support interactive terminal sessions.
- Improved node registry and RAG services for better agent performance.

fix(mesh/sync): accurately purge disjoint extra files on receiving end of a file manifest reconcile

fix(mesh/agent): ensure root path exists before watcher schedule to prevent Node stream crashes on purged re-attaches

feat(mesh): proactive session cleanup algorithm

feat(mesh): robust node connectivity and sync reliability fixes ...

- Fixed Node Registry persistence by adding missing db.commit() for online/offline status updates.
- Resolved 'Seed from Local' sync failures by correctly tracking mesh memberships in the TaskAssistant service.
- Enhanced frontend safety: disabled offline node selection and filtered source node candidates.
- Added 'Receiver Only' toggle for more granular control over multi-node synchronization.
- Corrected false-positive 'Source Node Disconnected' alert by verifying global connection status.
- Improved logging in Hub and Agent Nodes for better debugging of file transfer events.

feat: agent node mesh integration and UI polish

feat(mesh): implement Phase 6 Rich Health Metrics ...

- Enhanced agent-node to collect per-core CPU, frequency, available RAM, and load avg.
- Updated gRPC server to ingest and propagate rich health metrics via heartbeats.
- Updated Pydantic schemas to expose rich metrics in REST and WebSocket snapshots.
- Fixed UI tooltips in NodesPage.js to pop downwards and show detailed statistics.
- Fixed RegistrationResponse field mismatch (message -> error_message).
- Updated feature documentation for Agent Node Mesh.

Regenerate gRPC protos with task_id support

fix(grpc): fix relative imports in generated protos

feat(nodes): integrate live health into management tab and add unit-based hover tooltips

fix(fs): fix infinite loop in navigator and improve node error reporting

feat(fs): node-wide navigator, proactive sync, and polling for green dots

fix(fs): fix recursive creation/deletion and add sync status indicators

fix(agent): put agent on hub network and use service name for gRPC

fix(protos): regenerate Hub and Agent protos from project roots for absolute imports

chore(agent): remove duplicate proto files from root

fix(agent): sync proto missing task_id causing agent crash

fix: resolve 500 error in Node File Explorer by injecting orchestrator service and correlating task IDs in gRPC responses

feat: refine File System Navigator UI with terminal-like collapsible panel and folder icon toggle

Integrated FileSystemNavigator for Agent Node Mesh

feat(terminal): overhaul with xterm.js, raw TTY streaming, and dynamic resizing ...

- Replaced simulated terminal with xterm.js for native emulation
- Implemented raw TTY byte streaming over WebSockets
- Added support for persistent shell sessions and window resizing (ioctl TIOCSWINSZ)
- Improved SkillManager with a transparent TTY bypass to prevent thread pool saturation
- Added periodic heartbeat-based latency monitoring (RTT)
- Fixed skill_config parsing crash in AI Hub registry
- Enabled auto-scrolling and terminal-to-PTY grid synchronization

feat: implement real-time node monitoring and workspace initialization strategies