Auth seeding (last-resort guard)
- `_seed_builtin_admin()` in get_db() now skips when
VACUUM_WALL_SEED_BUILTIN_ADMIN=0 or when the users table already
contains any user — previously a fresh service start after a non-default
bootstrap (e.g. --mgmt-user alice) seeded a hard-coded `admin` with an
unrecoverable random password, shadowing the operator's account
- bootstrap_auth.py sets VACUUM_WALL_SEED_BUILTIN_ADMIN=0: bootstrap
creates the operator user itself on a fresh install, so exactly one
account exists and no seeded admin can appear
Frontend (session recovery)
- on page load/restore the in-memory TTL timer is gone, so a valid
7-day refresh token could sit in sessionStorage while the access token
is already expired server-side: the session `check` now attempts
exactly one refresh (POST /api/auth/refresh with the stored refresh
token) on 401 before treating the session as dead
- extract shared `_doRefresh()` used by both the `check` 401 fallback and
the `refresh` action (removes the duplicated rotation logic)
Tests
- update seeding tests to the new any-user-present check; add
test_seed_skipped_when_users_exist, test_seed_skipped_via_env,
test_bootstrap_flow_creates_exactly_one_user, and the auth-model JS
test suite (tests/test-auth-model.js)
Docs
- AGENTS.md: document VACUUM_WALL_SEED_BUILTIN_ADMIN
- architecture.md / hoover.md / security.md: describe the bootstrap
check 401 → one-refresh fallback path
Refresh/logout and token robustness
- drop the post-rotation refresh_tokens row delete in auth_refresh so
logout blacklists the current (rotated) refresh token; remove the
dead _clear_refresh_token_after_rotation helper and clear_active_refresh_token
- reject non-object JWT payloads in _extract_unverified_sub so crafted
Authorization headers return 401 instead of crashing with 500
SQLite user store
- make builtin-admin seeding idempotent: on a concurrent first start the
losing seeder re-checks, finds the winner, and returns instead of
raising IntegrityError
- per-thread sqlite connections + busy_timeout so Flask worker threads
don't hit cross-thread ProgrammingError / SQLITE_BUSY
- add LogsDirectory + /var/log/vacuum-wall to ReadWritePaths in both
systemd units so the fallback admin password actually lands on disk
Frontend
- skip apiFetch 401-recovery for public auth endpoints so a failed
login no longer logs out a valid session
- add /passkeys to the nav (passkey registration was unreachable);
remove the dead checkWebAuthnCapable export
- drop the CSP-blocked inline WS-URL script and the
__WS_URL_PLACEHOLDER__ plumbing; the WS URL is derived from location
Daemon / WS
- parse Sec-WebSocket-Protocol manually (web.Request.get_subprotocols
does not exist in aiohttp 3.13); X-Auth-Token is a custom-nginx
fallback only — docstring and security docs corrected
Install / system
- bootstrap_auth.py is now idempotent: preserves existing auth config
and syncs the admin password on re-runs (new reset_password helper)
- WebUI server block renders auth_basic off (the UI is JWT-protected)
- install.sh chown/chmod skips .git to avoid git dubious-ownership
breakage
- tolerate unreadable /etc/wireguard during system import
Contracts / docs
- create_user returns 409 on duplicate username per docs/api.md
- correct docs/api.md response shapes, docs/security.md blacklist
cleanup wording + one-refresh-per-user caveat, stale WS-URL
references, and the .htpasswd description
Tests: +7 regression tests (rotation/logout revocation, crafted-token
401, concurrent seeding); placeholder-substitution tests replaced with
serve-as-is SPA root tests.
Add require_session parameter to validate_token to enforce session_id
matching for refresh operations. Attacker with stolen refresh token can
no longer bypass session binding by omitting session_id from request.
Also adds backend guard against deleting builtin admin user (was only
blocked at Flask blueprint layer), and removes unused _ALL_RW variable.
- Add session_id to refresh tokens and enforce it during validation,
preventing stolen refresh tokens from being usable without the
originating browser session
- Set router.isAuthenticated via auth:login event after successful
login (previously only set at page load)
- Add console.warn logging to WS message parse/handler errors
- Improve _refreshPromise error handling in token refresh flow
- Document rate limiter in-memory limitation and CSP connect-src
same-origin requirement
- Add 3 tests for session-bound refresh token validation
- Make session_id parameter optional in validate_token — only enforced when
provided, allowing WebSocket auth which cannot carry custom headers
- Remove decode_token round-trip from daemon WS handler
- Override WebAuthn registration username from JWT user_ctx to prevent users
from registering credentials under another user's account
- Frontend no longer sends username for WebAuthn registration
- Redirect to login on 401 after token refresh fails for POST requests
- Remove redundant auth session check from login page load
- Add tests for session_id semantics and WebAuthn ownership guard
Session binding was bypassable: if the X-Session-Id header was absent,
validate_token skipped the check entirely, allowing a stolen JWT to be
used without the originating session.
Server-side: reject 401 early in Flask middleware and daemon WebSocket
handler when X-Session-Id is missing, before calling validate_token.
Updated validate_token to always enforce session_id matching for access
tokens (refresh tokens are unaffected as they carry no session_id claim).
Frontend: removed dead if (stored.session_id) guards in api.js since
the header is now always required. Added X-Session-Id to logout request
headers and always store session_id on login/refresh.
- Rate limiter tracks failures only; success resets counter
- Record failures/successes after password verification, not before
- WebAuthn rp_id/origin resolved dynamically from request domain
- Management domains auto-discovered from nginx backend config
- All WebAuthn operations validate domain against management list
- Add GET /api/auth/webauthn/capable endpoint for frontend checks
- Frontend checkWebAuthnCapable() function for domain-gated UI
- Timing side-channel fix: pre-compute dummy hash at module load
- Builtin admin seeded with random password (logged at WARNING)
- Logout handler returns consistent response shape
Apply ruff line-wrapping formatting to docs and test files.
Clarify auth middleware: extract user_permissions once before
subsystem check, removing conditional variable scoping.
- WireGuard: refactor to multi-interface 'access classes' model; extract config
generation and helpers into lib/wireguard.py; add per-class up/down endpoints
and API routes; update UI with class management pages and QR code component
- Firewall: fix zone creation with --new-zone before --set-target; skip
masquerade on public zone; add masquerade propagation for nftables backend
so NAT works when internal zones exit via public
- Network: rename sync event subsystem 'network' -> 'networkd'; always stamp
config hash even when deployment fails (fixes pending-changes detection)
- DHCP: add new API endpoint and update frontend page
- State/Sync: update state collectors and sync buses for new subsystems
- Docs: update API and config documentation for new endpoints and schemas
Add get_interface_ip() helper to resolve an interface's IPv4 address
via 'ip -o addr show'. Use it to back-propagate gateway into DHCP
ranges so clients receive their default route.
- set_dhcp_range() resolves gateway: explicit > existing range > iface IP
- DnsToFirewallSync and FirewallToDhcpSync sync ensure gateways are set
- Remove automatic masquerade toggle from DnsToFirewallSync
- Fix dnsmasq lease file path to /var/lib/misc/dnsmasq.leases
- Rename lease state field expires_at -> expires (ISO string)
- Add 'ip -o addr show' to sudo whitelist
- lib/system_import: new module to import system configs into JSON at daemon startup
- daemon/server.py: call import_all() during startup for config reconciliation
- daemon/handlers/nginx.py: simplify add_domain auth handling, remove duplicate code
- scripts/install.sh: replace inline Python setup with curl-based daemon API calls; apply IP forwarding at runtime
- hoover: bump internal asset versions to v=8
- pages: bump asset versions to v=9
- update-vendor.sh now creates webui/vendor symlinks (htm.js)
- install.sh calls update-vendor.sh after package install
- Add vendor/.empty and webui/vendor/.empty as directory placeholders in git
Add EventBus with loop guards to keep firewall, dnsmasq, wireguard,
and network configs consistent. Handlers emit SyncEvent after mutations;
subscribers compute diffs and write JSON without manual cascade loops.
- Replace per-zone --list-all calls with single --list-all-zones in _collect_firewall
- Add _parse_all_zones_output() parser with rich rules/rich-rules normalization
- Convert daemon shutdown to async with proper runner cleanup and socket unlink
- Add TimeoutStopSec=15 to vacuum-walld.service for graceful stop
- Fix exception handling in _collect_dnsmasq
- Remove management badge from proxy path rows
Replace the legacy top-level management key with a unified paths-based
model. Each domain now contains a paths map where each entry defines its
own backend, auth, headers, and flags (is_management, is_websocket).
- Add _migrate_config() to auto-migrate legacy formats on first load
- Remove set_management_proxy() and POST_NGINX_MANAGEMENT endpoint
- Update server_block.conf template to iterate paths with per-location auth
- Update daemon handler, API blueprint, state collector, and install script
- Add server config generation tests for paths, WebSocket, auth inheritance
- Update frontend proxy page to display per-path rows with flags
- Add find_cert_dir() to resolve both RSA and ECC (domain_ecc/) cert dirs
- Copy acme deploy hook to /deploy/ where acme.sh resolves it
- _parse_account_conf checks both legacy .account.conf and declarative config
- Skip public DNS check when all local IPs are private (NAT)
- Improve check message strings for validity and expiry status
- Support timezone-aware date formats in _days_until parsing
- Filter out "no" SAN domains in cert listing
- Bump frontend asset version cache keys
- Fix DOMContentLoaded race condition in app.js boot
- Fix spread operator in certs.js modal template
- Move acme.sh utilities (_run_acme, _find_acme, etc.) from lib/state to lib/acme
- Rewrite _parse_list_output to support pipe, tab, and column-separated formats
- Add ConflictError (409) to block issuing when cert already exists
- Move _find_issuance helper to detect in-progress issuance per domain
- Update issue_cert to check existing certs and return issuance status
- Fix start_polling to accept event loop explicitly
- Add sudoers entry for chown on vacuum-wall.conf
- Extend systemd ReadWritePaths for /run/nginx.pid and /var/log/nginx
- Update frontend to handle 'existing' issuance status
The SPA uses hash-based routing, so the catch-all route was dead code.
Replace with explicit routes for / and /vendor/<path>, plus a 404
handler that returns JSON for /api/ paths and empty HTML otherwise.
- lib/state.py: per-subsystem collectors with versioned state store
- daemon/server.py: state refresh on request, batch routing updates
- webui/static/hoover/html.js: new html tag template helper via htm.js
- webui/static/hoover/websocket.js: real-time state change notifications
- webui/static/hoover/vdom.js: VDOM improvements for keyed diff
- All frontend pages refactored to use html templates
- Add tests for state management and polling
- Update docs and AGENTS.md
- Register, view, and deactivate ACME accounts via API and UI
- 16-check validation framework for certificate issuance readiness
- DNS resolution, port, nginx, and firewall pre-flight checks
- External IP detection with NAT support and fallback providers
- Account card and settings modal in certificates page
- Guard certificate issuance behind account registration
- Update modal CSS to overlay-based approach
- 1000+ lines of tests for validation and account handlers
Phase 1-4: Networkd subsystem
- lib/network.py: systemd-networkd config renderer (.network INI files)
with full schema support: [Match], [Link], [Network], [Address], [Route],
[DHCPv4], [DHCPv6] sections. One Address/=DNS= line per value per spec.
Route sections use #N suffix per systemd.syntax(7).
- lib/network.py: generate_network_files() with 50-<name>.network prefix
and stale file cleanup
- lib/network.py: collect_upstream_dns() filters local/private DNS
- lib/network.py: infer_dhcp_ranges() and infer_zones() helpers
- daemon/handlers/network.py: routes for GET/POST /network/interfaces
and full apply with DNS upstream sync to dnsmasq
- webui/api/network.py: Flask blueprint for /api/network/* endpoints
- webui/api: interfaces page updated with IP config inline editing
- lib/state.py: networkd collector using parse_networkctl_status()
- system/sudoers.d/vacuum-walld: networkctl + systemd-network rules
- system/systemd/vacuum-walld.service: ReadWritePaths for /etc/systemd/network
- install.sh: ACME email now optional, configured from WebUI
- lib/acme.py: get_email() falls back to declarative config
Phase 5: Code review fixes
- daemon/server.py: path params now win over JSON body and query params
in request body merge (prevents config save name override)
- daemon/server.py: remove dead 'import re'
- daemon/handlers/network.py: replace Path.mkdir() with sudo mkdir
for /etc/systemd/network (ProtectSystem=strict compatibility)
- system/sudoers.d/vacuum-walld: pin systemctl to specific commands
(reload/is-active dnsmasq instead of wildcard)
- system/sudoers.d/vacuum-walld: restore !requiretty and section comment
- lib/network.py: remove unused _MANAGEMENT_PORTS constant
- webui/api/network.py: remove redundant body[\name\] = name in save_interface
Tests: 332 passing (110 new/updated), ruff clean
- Add daemon/ module with aiohttp server, sync client, and handler registry
- Add daemon/handlers/ for privileged operations (acme, dnsmasq, firewall, logs, nginx, wireguard)
- Add system/acme-deploy.py, vacuum-walld sudoers and systemd service
- Update API routes to use daemon client instead of lib/ directly
- Update lib/, tests/, and webui/ for new architecture
- Update docs and deployment scripts
- Fix WireGuard private key leak in API responses and config updates
- Update systemd service to serve from repo root with adjusted sandbox
- Add CLI flags, idempotency, and dev mode to install.sh
- Extract common utilities to lib/common.py and webui/api/common.py
- Migrate frontend to htmx for simpler, more maintainable UI
- Update docs to reflect current architecture and deployment model
- Vendor htmx dependencies per project requirements
- Add deploy(domain) to acme.renew() to register deploy hook after
renewal, matching the pattern in issue()
- Hoist run and run_proc imports to module level in wireguard,
removing 4 inline imports for consistency
- wireguard: POST /peers with JSON encoding (was /add-peer)
- rules: delete by rule_id in URL path (was JSON body); pass rule objects with id from server; add hx-disable to initial render
- nat: port forward delete uses URL path params to match blueprint
- nat: masquerade toggle uses native hx-post/hx-vals (was inline fetch)
- app.js renderers updated to use URL path deletes for rules and forwards
- remove TODO.md
New two-step config flow: POST /config saves desired state to
config/firewall/config.json, GET /config/pending diffs against live
firewalld state, POST /config/apply synchronizes live state. Adds target
normalization helpers and full test coverage for config CRUD and pending
diff logic.
Replace all hardcoded /home/wall/vacuum-wall paths in lib/ with Path(__file__).resolve()
auto-discovery. Move config files from data/ to config/<subsystem>/config.json.
ACME now uses ACME_HOME env var and data/acme/ for cert storage. Systemd units
and sudoers use {{ USER_NAME }}, {{ PROJECT_DIR }}, {{ ACME_HOME }} Jinja2
template variables for install-time substitution. Remove sys.path.insert boot
strap from test files.
- dashboard.html: Fix zones, leases, wg_status, cert key names, add services var
- server.py: Pass services to dashboard template via _get_service_status()
- lib/acme.py: Fix dead third date format (%Y%m%d%H%M%z) using astimezone(UTC)
- lib/wireguard.py: Add -- separator to cp command to match sudoers rule
- lib/nginx.py: Replace shallow dict.copy() with {**...} for DEFAULT_SSL
- AGENTS.md: Update test count 149 -> 154
- docs/api.md: Rename cert field expiry -> expires_at
mteehan