Freeze all leading system prompt blocks per conversation, including the main prompt, skills, workspace context, recent conversations, personalization, AGENTS.md, custom system prompts, and disabled-tool notices. Tool schemas remain dynamically generated outside the frozen prompt path.
Remove obsolete prompt injection files and code paths for thinking mode, deep thinking mode, sub-agent guidelines, and todo guidelines so prompt assembly stays simpler and cache-friendly.
Add personalization-controlled conversation_continuity with high/medium/low levels, rename the UI copy to 对话连续性, and wire it through backend sanitization, prompt generation, manage_personalization, frontend settings, and tool result rendering.
Add auto communication style support via prompts/auto_style.txt, matching the existing separate-file pattern used by human_like_style.txt.
Implement Docker Web project management on top of workspace_id while keeping host mode workspace behavior intact.
Backend changes:
- Move Docker user files to users/<user>/projects/<project_id>/{project,data,logs} and shared user state to users/<user>/personal.
- Add automatic legacy migration from old project/data/logs/agentskills layouts and from the temporary project/shared layout.
- Generate project IDs server-side from the project name, reserve internal IDs, and hide project paths from the UI.
- Add project rename/delete APIs; deleting a Docker project removes its project directory and releases terminal/container state.
- Add host workspace rename/delete APIs; deleting a host workspace only removes catalog entries and keeps files on disk.
- Scope Docker containers, terminals, uploads, conversations, storage, and running task metadata by workspace_id.
- Store private skills under users/<user>/personal/agentskills and sync them into each project workspace.
- Inject the current project/workspace name through prompts/workspace_system.txt.
Frontend changes:
- Replace Docker file-tree area with project management UX and reuse host multi-workspace running-task behavior.
- Add a management dialog for creating, renaming, and deleting projects/workspaces.
- Show project labels instead of internal IDs in toasts and project lists.
- Clear stale conversation history during project/workspace switching and show loading until the new list is ready.
- Keep running-task loader/check behavior consistent across host workspaces and Docker projects.
Validation:
- python3 -m py_compile modules/user_manager.py modules/host_workspace_manager.py modules/skills_manager.py core/main_terminal_parts/tools_execution.py server/auth.py server/context.py server/status.py server/conversation.py server/tasks.py utils/context_manager.py
- python3 -m unittest test.test_server_refactor_smoke
- npm run build --silent 2>&1 | tail -n 20
Refactor conversation persistence into workspace-scoped storage with host workspace legacy migration, current workspace filtering, and active host workspace state protection.
Add proactive recent-conversation prompt injection with personal-space controls, configurable item count, empty-conversation filtering, and frozen per-conversation prompt metadata to keep request caches stable.
Introduce conversation_search and conversation_review tools for current-workspace history: multi-keyword search, list mode, current conversation exclusion, message/tool counts, formatted model output, read/save review modes, long-review fallback, and .agents/review storage.
Move internal workspace artifacts under .agents, including uploads, skills, compact results, and reviews; update prompt file-tree visibility, compression output, upload paths, and hidden versioning exclusions.
Add create_skill workflow and a Skills tool category, move read_skill into Skills, validate/归档 skill folders, support host global skills and Docker/API private skills, and sync private skills into .agents/skills and skill prompts.
Update frontend tool icons, statuses, and expanded renderers for conversation recall and skills workflows; add workspace conversation and skills-manager tests.
This commit consolidates a large set of host-mode security and UX fixes across backend and frontend.
Security / execution model:
- Make permission/execution mode switches apply immediately (no deferred pending-only behavior).
- Keep runtime mode change notices durable and visible via user-message guidance path.
- Add explicit runtime message source taxonomy support and persistence:
user / presend / guidance / notify / sub_agent / background_command.
- Ensure guidance/notify insertion is batched per tool-cycle so mid-run inserts are committed together.
- Separate notify from guidance semantics for mode-change notices.
- Add direct-mode auto-fallback handling visibility on frontend.
Approval and command gating:
- Align run_command approval pre-check logic with docker behavior for approval/auto_approval modes in host execution paths.
- Keep approval retry flow and final-result semantics intact when permission errors trigger approval.
- Externalize forbidden command keyword list into JSON config:
config/forbidden_commands.json.
- Update forbidden command rejection message to user-facing wording:
用户不允许执行包含“... ”的指令.
Prompt/runtime guidance:
- Refine execution-mode runtime rule text to avoid endless workaround loops:
one safe alternative first, then request switching to direct mode when truly required.
Chat rendering / message UX:
- Prevent injected guidance/notify user-messages from incorrectly triggering assistant work header/timer chain.
- Restore correct visibility after history reload while preserving runtime-source behavior.
- Update user-header icon/label rendering by message source:
- guidance -> 引导 + navigation icon
- notify/sub_agent/background_command -> 通知 + bell icon
- user/presend -> keep normal user header.
- Add new icons: static/icons/navigation.svg, static/icons/bell.svg.
Approval panel UX:
- Convert desktop approval panel from layout-compress sidebar to overlay + blur sheet.
- Fix panel transition behavior (remove unintended vertical motion; use dedicated desktop overlay transition).
- Improve approval result display:
- decision and reason split lines,
- green/red status styling,
- multiline reason rendering,
- auto-collapse delay adjusted to 3s.
Execution mode TTL feedback:
- When direct mode auto-expires to sandbox, frontend menu state is updated from status snapshot.
- Show non-auto-dismiss warning toast to explicitly notify auto-fallback.
Misc consistency updates:
- Execution mode label text cleanup in composer.
- Runtime notice wording normalization (权限模式修改为..., 执行环境修改为...).
Build/test notes:
- Frontend rebuilt after UI changes.
- Smoke tests passed for server refactor path.
- Add docker risk markers config (config/docker_risk_markers.json) and load it in tools permission evaluation.
- In docker mode, pre-evaluate run_command/terminal_input: readonly-safe and no-risk -> direct run; otherwise require approval/auto-approval.
- Keep docker write_file/edit_file behavior aligned with agreed policy: approval requires review, auto_approval passes directly.
- Pass risk markers into auto approval payload and include current working directory context for approval agent.
- Fix permission dropdown layout: only use split two-column menu when execution mode section is enabled; remove empty right blank area in docker mode.
- Center mobile approval overlay to prevent sheet overflow on small screens.
This commit lands a broad host-security architecture update focused on enforceable sandbox execution, clearer permission semantics, and operator usability.
Core execution/security changes
- Introduce and wire a unified host sandbox runner for multi-OS execution:
- macOS: sandbox-exec
- Linux: bubblewrap + seccomp
- Windows: WSL2 path
- Remove silent fallback behavior in failure paths; sandbox-unavailable cases now fail closed instead of dropping to unsafe host execution.
- Ensure host execution mode (sandbox/direct) is propagated consistently into runtime components, including sub-agent startup.
Permission mode model upgrade
- Rework readonly/approval behavior for run_command from brittle command-text gating to execution-layer enforcement:
- readonly: run_command executes in read-only sandbox profile.
- approval: run_command first executes read-only; when permission-denied is detected, request approval and retry once with writable sandbox for that single call.
- Tool loop now returns final post-approval execution result, not intermediate permission-denied payloads.
- Update permission-mode system messaging to describe user-visible behavior without exposing internal implementation details.
Path authorization system
- Add dynamic host policy module and persisted policy file.
- Support dual path classes:
- writable_paths (read+write)
- readable_extra_paths (read-only)
- Enforce file access in file_manager by access type (read vs write) under host mode.
- Add host-only frontend 路径授权 management dialog (settings三级菜单入口), including mode switch between 可读可写 and 仅可读 with separate drafts and save flow.
Sub-agent and terminal alignment
- Sub-agent process launch now respects host execution mode and sandbox controls in host mode.
- Keep terminal session model compatible with sandbox-first behavior and execution-mode propagation.
Tool surface updates
- Remove legacy file-management tools from active tool definitions (create_file/create_folder/rename_file/delete_file) while preserving historical conversation compatibility.
Docs updates
- Add dedicated architecture doc: docs/host_sandbox_and_permission_model.md
- Refresh README and AGENTS sections to reflect updated permission/execution model and path-authorization semantics.
Validation performed
- python unittest smoke suite passes: test.test_server_refactor_smoke
- frontend build passes: npm run build
- syntax checks for touched Python modules completed
This commit introduces a substantial security and runtime architecture update for host mode, with three major goals: (1) enforce OS-level sandboxing by default, (2) support a controlled temporary direct-execution escape hatch for admin users, and (3) eliminate silent fallback behavior that could hide risk.
Backend/runtime changes:\n- Added a unified host sandbox runner (macOS sandbox-exec / Linux bubblewrap+seccomp / Windows WSL2).\n- Converted host terminal creation to sandboxed interactive shells by default.\n- Kept run_command/run_python on the same execution policy and added host execution mode plumbing (sandbox|direct).\n- Added session-scoped direct mode with TTL auto-revert (default 10 minutes), configurable via environment.\n- Added execution mode APIs (GET/POST /api/execution-mode), host+admin gating, status propagation, and rate limiting.\n- Added runtime refresh hooks so tool calls honor TTL expiration and mode transitions consistently.\n- Removed docker->host silent fallback paths: docker startup/runtime failures now fail fast instead of degrading silently.\n- Added host sandbox prompt injection (host-only) to guide agent behavior under permission constraints.
Configuration and policy changes:\n- Added HOST_EXECUTION_MODE_DEFAULT and HOST_EXECUTION_DIRECT_TTL_SECONDS.\n- Added HOST_SANDBOX_MACOS_WRITABLE_PATHS to support user-defined writable path allowlists.\n- Updated macOS sandbox profile generation to use minimal defaults (workspace + tmp + /dev/null) plus explicit allowlist extensions.\n- Updated .env.example documentation for new execution/sandbox controls.
Frontend changes:\n- Extended permission popover to a two-column model (Permission + Execution Environment) for host admin sessions.\n- Added execution mode state/options to app state, fetching, and status synchronization flows.\n- Added execution mode switching action and user feedback toasts.\n- Kept warning emphasis via text styling only (removed warning border per UX request).
Validation:\n- Python syntax checks passed for modified backend modules.\n- Existing smoke tests passed: python3 -m unittest test.test_server_refactor_smoke\n- Frontend production build passed: npm run build
