Three related bugs fixed:
1. Frontend xterm.js crash: TerminalPage rendered ALL sessions with display:none
for inactive ones. xterm.js crashes when initialized in a hidden container
(Viewport can't read dimensions). Fix: only render the active session's
TerminalComponent using conditional rendering.
2. Backend websocket disconnect cascade: When client disconnected (due to #1),
the server tried to send 'connected' status on dead socket, caught the
WebSocketDisconnect in a generic except block, then tried to close() again
causing RuntimeError. Fix: catch WebSocketDisconnect specifically and suppress
close() errors.
3. Stale DB sessions: After server restart, DB still had old terminal session
rows but no in-memory sessions. list_terminal_sessions returned these ghosts,
causing the frontend to render dead tabs. Fix: skip DB-only sessions that
have no live in-memory counterpart.
Quality gates: pytest (15/15 passed), tsc clean, vitest (7/7 passed)
The frontend router navigates to /instances/:instanceId/terminal without
project_id or repo_id. The backend terminal REST endpoints were requiring
these path params, causing 404s.
- Simplify _get_terminal_instance to validate by instance_id only
- Update all REST routes from /projects/{pid}/repositories/{rid}/instances/{iid}/terminal/*
to /instances/{instance_id}/terminal/*
- Update frontend API client to match new paths
- Update useTerminalSessions hook to take instanceId only
- Update TerminalPage to use simplified hook
- Update tests to match new paths
Fixes: 404 on GET /projects/repositories/instances/{id}/terminal/sessions
- Add test_tool_instances_legacy.py with 8 unit tests:
- dockerfile definition type builds from template
- dockerfile build failure raises HTTP 500
- compose definition type renders template
- manifest compiler is NOT called for legacy types
- start_instance legacy/compose/dockerfile types all skip manifest flow
- start_instance manifest type correctly invokes compiler
- Mark T3.2 and T3.3 tasks complete in OpenSpec
- Add openspec/docs/tool-workshop-guide.md with user guide covering
definition types, manifest creation workflow, base definitions,
migration path, and permissions
The production database was stamped with a migration that no longer exists
in the codebase (created on another branch, applied, then removed). This
adds a no-op bridge migration so Alembic can reconcile the DB state.
- Create bridge migration 2026_05_28_add_tool_definition_manifests (no-op)
- Re-chain terminal_sessions migration to depend on the bridge
- Fixes startup failure: Can't locate revision identified by ...
- Add tool_definitions API client with types for manifests
- Add ManifestEditor component: base image selector, package editors
(apt/npm/pip/node), script editors (build/startup), mount schema
designer, runtime config, and live preview panel
- Integrate ManifestEditor into Tool Workshop as 'Manifest (Declarative)'
definition type alongside Compose and Dockerfile
- Update ToolType API types to include manifest_id and 'manifest'
definition_type
- Frontend builds clean, TypeScript typecheck passes
- Add ToolDefinitionManifest model with base image versioning
- Add manifest compiler: Dockerfile + Compose generation from JSON manifests
- Add permission fixer: post-start chown/chmod for mount policies
- Add tool definition CRUD API with live compile preview endpoint
- Integrate manifest-based startup flow in start_instance
- Add Alembic migration with data conversion for pi-agent
- Add 48 unit tests for manifest compiler, permission fixer, docker service
- Keep backward compatibility with legacy dockerfile_template/compose_template
Migration: applied successfully. Pi-agent converted to manifest.
Quality gates: pytest (146 passed, 4 pre-existing unrelated failures)
- Add WebSocket route /ws/tool-instances/{instance_id}/terminal/{session_id}
- Preserve /terminal as default-session alias for backward compatibility
- Extract shared _handle_terminal_websocket handler for both routes
- Add REST endpoints: GET list, POST create, DELETE close, POST reset, POST rename
- Preserve legacy POST .../terminal/reset as default session alias
- Add frontend API client (apps/web/src/api/terminal.ts)
- Add useTerminalSessions React hook for session CRUD + state management
- Add integration tests for auth requirements on all new endpoints
Quality gates: pytest (8 new passed, 182 total passed, 51 pre-existing failures)
- Add TerminalSessionModel DB table with instance_id FK, name, status,
created_at, last_activity_at, closed_at columns
- Add Alembic migration for terminal_sessions table
- Refactor TerminalManager to use composite key (instance_id, session_id)
supporting up to 5 concurrent sessions per instance
- Add create_session, get_session, get_sessions_for_instance, close_session
- Preserve get_or_create_session for backward compatibility (default session)
- Fix attach_websocket to only close sockets within same session
- Add name (auto-generated 'Session N') and status tracking to TerminalSession
- Add 7 unit tests for multi-session logic
Quality gates: pytest (7 new passed, 174 total passed, 51 pre-existing failures)
- get_container_id() and get_container_name() now lowercase the
instance name before passing to docker ps --filter, because
Docker container names are lowercase internally and the filter
is case-sensitive. This caused container_id to never be captured
when instance.name contained uppercase chars (e.g. 'Headquarter'),
breaking terminal WebSocket connections.
- Also guard proc.stdout being None in start_cloudflared_tunnel().
- Add unit tests for get_container_id and get_container_name.
Quality gates: pytest (14 passed), python clean
- Add stdin_open: true and tty: true to dockerfile-based compose generation.
Without these, bash (PID 1) exits immediately, causing a container restart
loop that makes the instance invisible to docker ps and triggers 4004.
- Treat WebSocket close codes 4001/4003/4004 as permanent errors in the
frontend. Stop retrying and show the server reason to the user.
- Prevent visibilitychange handler from resetting retry attempts after a
permanent error has occurred.
- Use docker ps -a in get_container_id/get_container_name to find
stopped/exited containers for diagnostics.
Quality gates: tsc --noEmit (pass), pytest (98 passed, 4 pre-existing failures)
- Fix git mount clone to check correct path (repo-clone subdir)
- Pull updates instead of re-cloning when git mount dir exists
- Add compose file sanitization to remove invalid port 0 mappings
- Fixes startup failures for existing instances with old compose files
- Adds pi-agent to tool_types table with terminal interface
- Includes Dockerfile template for pi.dev coding agent
- Idempotent: checks for existing entry before insert
- Remove absolute path requirement from target_path validation
- Resolve relative paths against working_directory at instance startup
- Fall back to /home/user if no working_directory is configured
- Update frontend to allow relative target paths
- Update spec to document relative path support
- Update tests to allow relative paths and test path traversal rejection
- _checkout_branch now returns bool and falls back gracefully on failure
- Glob warning message includes matched file count
- Fix database model comment to reference remote_url
- Update tests for new branch checkout behavior
All 51 tests pass
- Change git mount schema from repo_id to remote_url
- Remove database lookups for git mount resolution
- Clone directly from URL at instance startup
- Simplify frontend UI to text input for Git URL
- Fix route ordering in git_repositories.py to prevent 422 errors
- Update all tests to use remote_url field
Breaking change: Git mounts now use remote_url instead of repo_id
- Add POST /repositories endpoint for external repos (no project_id)
- Update GitRepositoryResponse to allow nullable project_id
- Update list_repositories to support listing all user repos
- Add _pull_repository_updates for auto-pull on container creation
- Update git mount validation to allow external repos
- Frontend: Update listRepositories to support optional projectId
- Spec updates: external repos, auto-clone, per-instance isolation
The server is missing 2026_05_27_make_project_id_nullable.py but the
merge migration referenced it. Removing the merge migration leaves a
clean single head chain.
The server has both the old (2026_05_27_make_project_id_nullable) and
new (2026_05_27_external_repos) migration files, creating two heads.
This merge migration resolves them into a single head.
- Make project_id nullable in git_repositories table (migration)
- Allow external repos not tied to any project
- Update validation to allow user-owned external repos in git mounts
- Add /projects/repositories endpoint to list all user repos
- Update frontend to fetch all user repos for git mount selector
- TypeScript and build pass
- Create shared_validators.py with validate_mount_path, validate_files, validate_env_vars, validate_volumes
- Refactor config_folders.py to use shared validators
- Refactor tool_configs.py to use shared validators
- Refactor config_profiles.py to use shared env_vars validator
- Reduce ~80 lines of duplicate validation code
- Add startup_command field to ToolType model and API
- Execute startup command before interactive shell in terminal sessions
- Add tmux and ranger to OpenCode container spec
- Update Tool Workshop UI with startup_command input for terminal types
- Add backend tests for startup_command CRUD operations
- Sync specs: tool-terminal, tool-types-definition, opencode-web-server
- New spec: tool-terminal-startup-command
Quality gates: Frontend typecheck/lint passed. Backend tests blocked by environment (Python/Docker not available).
OpenSpec: terminal-startup-and-container-tools
Frontend:
- Clear xterm.js screen when receiving 'connected' status after reset
- Send resize message after clearing to ensure proper dimensions
- Fixes terminal artifacts after reset
Backend:
- Fix data loss bug: text starting with '{' but not valid JSON was silently dropped
- Now writes such text to session as regular input
- Fixes missing characters when user types '{'
Frontend:
- Fix term.onData to use wsRef.current instead of captured ws variable
- Fix fitTerminal to use wsRef.current for resize messages
- Fix sendData callback to use wsRef.current
- This fixes 'cannot type' after WebSocket reconnect
Backend:
- Add SessionRef class for mutable session reference
- Update _read_loop and _write_loop to use SessionRef
- Reset now updates session_ref.session instead of returning
- This keeps the WebSocket alive after reset instead of closing it
Instead of sending stty commands through the user's terminal session
(which causes 'inappropriate ioctl' errors), send SIGWINCH signal to
the docker exec process on the host. Docker exec should forward this
to the container process, causing the shell to re-read its terminal size.
This avoids:
- Visible stty commands in the terminal
- ioctl errors from stty
- Interference with user's shell session
Reverted docker exec back to -it (required for interactive bash).
Instead, sends stty command with \r to hide it from the terminal display:
- \r moves cursor to start of line (overwrites prompt)
- stty command executes silently (no output on success)
- \r moves cursor back to start, hiding echoed command
This sends stty on EVERY resize so the container shell always matches
frontend dimensions.
Removes -t flag from docker exec so it uses our PTY slave directly instead
of creating its own PTY inside the container. This allows TIOCSWINSZ on the
host PTY master to propagate naturally to the container shell via SIGWINCH.
Also removes all stty command injection logic since resize now works natively.
When FastAPI parses the request body and model_dump() is called,
nested MountItem models are already serialized to plain dicts.
The update handler was unconditionally calling model_dump() again,
causing AttributeError on dict objects.
Fix: Check if mount items are already dicts before calling model_dump().
Fixes: 422 error when updating profiles with mounts.
Backend:
- Remove _stty_sent guard to send stty on EVERY resize
- Use stty -echo to hide command, then delete the command line with ANSI escapes
- Change log level from info to debug
Frontend:
- Add window resize listener as fallback to ResizeObserver
- 250ms debounce to avoid excessive refits
- Proper cleanup on unmount
alex
Alex Blank