- Rename compose profile 'receiver' -> 'observer' across all docs
- Rename docker-compose service 'packet-capture' -> 'observer' and volume
'packetcapture_data' -> 'observer_data'
- Change COMPOSE_PROJECT_NAME fallback from 'hub-dev' to 'hub' across all
compose files, Makefile, .env, docs
- Remove legacy interface-mock service from test compose file
Simplify the variable name to remove the legacy LetsMesh decoder prefix.
Also fix unparenthesized except tuples in web/app.py and promote the
parenthesized-exception rule to a prominent position in AGENTS.md.
Replace subprocess-based packet decoding with native Python meshcoredecoder>=0.3.2.
The decoder is now always enabled (no toggle), removing ENABLED/COMMAND/TIMEOUT
config vars. Adds _enrich_payload_decoded() to compensate for payload classes that
lack to_dict() overrides in the library. Removes Node.js/npm from Dockerfile and
deletes the patches/ directory. Adds parenthesized exception rule to AGENTS.md.
Remove the meshcore_interface component in favor of external
meshcore-packet-capture for data ingestion. Rename receiver_node_id
to observer_node_id across all models, schemas, handlers, and API
routes. Add Alembic migration for the column/table renames. Fix
frontend JS property name mismatch that prevented the Receiver column
from displaying observer data.
Pin starlette<1.0.0 to avoid TemplateResponse breaking change, update
the TemplateResponse call to new-style request-first signature, and
bump Python version to 3.14 across Dockerfile, pyproject.toml, and
pre-commit hooks.
Update path hash handling to accept variable-length hex-encoded hashes
(e.g. "4a" for single-byte, "b3fa" for multibyte) instead of requiring
exactly 2-character hashes. Bump meshcore dependency to >=2.3.0.
- Update normalizer to accept even-length hex strings >= 2 chars
- Update schemas and model docstrings for variable-length hashes
- Add tests for multibyte and mixed-length path hash round-trips
- Fix web test flakiness from local .env datetime locale leaking
- Use hmac.compare_digest for constant-time API key comparison in auth
and metrics endpoints to prevent timing attacks
- Escape user-controlled data in admin JS templates (members, node-tags)
to prevent XSS via innerHTML
- Escape </script> sequences in embedded JSON config to prevent XSS
breakout from <script> blocks
- Add configurable WEB_TRUSTED_PROXY_HOSTS setting instead of trusting
all proxy headers unconditionally
- Warn on startup when admin is enabled with default trust-all proxy
- Remove legacy HTML dashboard endpoint (unused, superseded by SPA)
- Add comprehensive auth and dashboard test coverage
Joins NodeTag (key='role') to the node last seen Prometheus metric so
alert rules can target infrastructure nodes only (role="infra").
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add /metrics endpoint with Prometheus gauges for nodes, messages,
advertisements, telemetry, trace paths, events, and members. Include
per-node last_seen timestamps for alerting. Add Alertmanager service
to Docker Compose metrics profile with default blackhole receiver.
Add NodeNotSeen alert rule (48h threshold). Add 1h time window to
all windowed metrics alongside existing 24h/7d/30d windows.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Implement cache-control middleware to optimize browser caching and reduce
bandwidth usage. Static files are cached for 1 year when accessed with
version parameters, while dynamic content is never cached.
Changes:
- Add CacheControlMiddleware with path-based caching logic
- Register middleware in web app after ProxyHeadersMiddleware
- Add version query parameters to CSS, JS, and app.js references
- Create comprehensive test suite (20 tests) for all cache behaviors
Cache strategy:
- Static files with ?v=X.Y.Z: 1 year (immutable)
- Static files without version: 1 hour (fallback)
- SPA shell HTML: no-cache (dynamic config)
- Health endpoints: no-cache, no-store (always fresh)
- Map data: 5 minutes (location updates)
- Custom pages: 1 hour (stable markdown)
- API proxy: pass-through (backend controls)
All 458 tests passing, 95% middleware coverage.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
## i18n Refactoring
- Refactor admin translations to use common composable patterns
- Add common patterns: delete_entity_confirm, entity_added_success, move_entity_to_another_node, etc.
- Remove 18 duplicate keys from admin_members and admin_node_tags sections
- Update all admin JavaScript files to use new common patterns with dynamic entity composition
- Fix label consistency: rename first_seen to first_seen_label to match naming convention
## Translation Documentation
- Create comprehensive translation reference guide (languages.md) with 200+ documented keys
- Add translation architecture documentation to AGENTS.md with examples and best practices
- Add "Help Translate" call-to-action section in README with link to translation guide
- Add i18n feature to README features list
## Documentation Audit
- Add undocumented config options: API_KEY, WEB_LOCALE, WEB_DOMAIN to README and .env.example
- Fix outdated CLI syntax: interface --mode receiver → interface receiver
- Update database migration commands to use CLI wrapper (meshcore-hub db) instead of direct alembic
- Add static/locales/ directory to project structure section
- Add i18n configuration (WEB_LOCALE, WEB_THEME) to docker-compose.yml
## Testing
- All 438 tests passing
- All pre-commit checks passing (black, flake8, mypy)
- Added tests for new common translation patterns
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Operators can now disable specific pages (Dashboard, Nodes, Advertisements,
Messages, Map, Members, Pages) via FEATURE_* environment variables. Disabled
features are fully hidden: removed from navigation, return 404 on routes,
and excluded from sitemap/robots.txt. Dashboard auto-disables when all of
Nodes/Advertisements/Messages are off. Map auto-disables when Nodes is off.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The admin pages only checked config.admin_enabled but not
config.is_authenticated, allowing unauthenticated users to access
admin functionality when WEB_ADMIN_ENABLED=true. Additionally, the
API proxy forwarded the service-level Bearer token on all requests
regardless of user authentication, granting full admin API access
to unauthenticated browsers.
Server-side: block POST/PUT/DELETE/PATCH through the API proxy when
admin is enabled and no X-Forwarded-User header is present.
Client-side: add is_authenticated check to all three admin pages,
showing a sign-in prompt instead of admin content.
https://claude.ai/code/session_01HYuz5XLjYZ6JaowWqz643A
Replace server-side rendered Jinja2 page routes with a client-side SPA
using ES modules, lit-html templating, and a custom History API router.
All page rendering now happens in the browser with efficient DOM diffing.
Key changes:
- Add SPA router, API client, shared components, and 14 page modules
- Serve single spa.html shell template with catch-all route
- Remove server-side page routes (web/routes/) and legacy JS files
- Add centralized OKLCH color palette in CSS custom properties
- Add colored nav icons, navbar spacing, and loading spinner
- Add canonical URL and SEO path exclusions to SPA router
- Update charts.js to read from shared color palette
- Update tests for SPA architecture (template-agnostic assertions)
- Update AGENTS.md and README.md with SPA documentation
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add GPS coordinate fallback: use tag coords, fall back to model coords
- Filter out nodes at (0, 0) coordinates (likely unset defaults)
- Add "Show" filter to toggle between All Nodes and Infrastructure Only
- Add "Show Labels" checkbox (labels hidden by default, appear on hover)
- Infrastructure nodes display network logo instead of emoji
- Add radius-based bounds filtering (20km) to prevent outlier zoom issues
- Position labels underneath pins, centered with transparent background
- Calculate and return infra_center for infrastructure node focus
- Initial map view focuses on infrastructure nodes when available
- Update popup button to outline style
- Add comprehensive tests for new functionality
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Replace PAGES_HOME with CONTENT_HOME configuration (default: ./content)
- Content directory now contains pages/ and media/ subdirectories
- Add support for custom logo at $CONTENT_HOME/media/images/logo.svg
- Custom logo replaces favicon and navbar/home logos when present
- Mount media directory as /media for serving custom assets
- Simplify default logo to generic WiFi-style radiating arcs
- Update documentation and example directory structure
- Update tests for new CONTENT_HOME structure
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Renamed network route, template, and tests to dashboard
- Added logo.svg for favicon and navbar branding
- Created reusable Jinja2 icon macros for navigation and UI elements
- Updated home page hero layout with centered content and larger logo
- Added Map button alongside Dashboard button in hero section
- Navigation menu items now display icons before labels
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Allows adding static content pages (About, FAQ, etc.) as markdown files
with YAML frontmatter. Pages are stored in PAGES_HOME directory (default:
./pages), automatically appear in navigation menu, and are included in
the sitemap.
- Add PageLoader class to parse markdown with frontmatter
- Add /pages/{slug} route for rendering custom pages
- Add PAGES_HOME config setting to WebSettings
- Add prose CSS styles for markdown content
- Add pages to navigation and sitemap
- Update docker-compose.yml with pages volume mount
- Add comprehensive tests for PageLoader and routes
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add /api/v1/nodes/prefix/{prefix} for prefix-based node lookup
- Change /api/v1/nodes/{public_key} to exact match only
- /n/{prefix} now simply redirects to /nodes/{prefix}
- /nodes/{key} resolves prefixes via API and redirects to full key
Allow users to navigate to a node using any prefix of its public key
instead of requiring the full 64-character key. If multiple nodes match
the prefix, the first one alphabetically is returned.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When users try to access /a/ without valid OAuth2Proxy headers (e.g.,
GitHub account not in org), they now see a friendly 403 page instead
of a 500 error. Added authentication checks to all admin routes.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Replace node type badge with icon in admin tag editor
- Add Edit/Add Tags button on node detail page (when admin enabled and authenticated)
- Remove automatic seed container startup to prevent overwriting user changes
- Remove unused 'coordinate' value type from node tags (only string, number, boolean remain)
- Fix XSS vulnerability by using data attributes instead of inline
onclick handlers in node_tags.html template
- Fix URL injection by using urlencode for all redirect URL parameters
- Add validation to reject moves where source and destination nodes
are the same (returns 400 Bad Request)
- Add error handling for response.json() calls that may fail
- Add missing test coverage for update endpoint error scenarios
Implement CRUD operations for NodeTags in the admin interface:
- Add NodeTagMove schema for moving tags between nodes
- Add PUT /nodes/{public_key}/tags/{key}/move API endpoint
- Add web routes at /a/node-tags for tag management
- Create admin templates with node selector and tag management UI
- Support editing, adding, moving, and deleting tags via API calls
- Add comprehensive tests for new functionality
The interface allows selecting a node from a dropdown, viewing its
tags, and performing all CRUD operations including moving a tag
to a different node without having to delete and recreate it.
- Add CONTACT_CLEANUP_ENABLED and CONTACT_CLEANUP_DAYS settings
- Implement remove_contact and schedule_remove_contact on device classes
- During contact sync, remove stale contacts from companion node
- Stale contacts (not advertised for > N days) not published to MQTT
- Update Python version to 3.13 across project config
- Remove brittle config tests that assumed default env values
Replace presentation-layer deduplication with collector-level approach:
- Add event_hash column to messages, advertisements, trace_paths, telemetry tables
- Handlers compute content hashes and skip duplicate events at insertion time
- Use 5-minute time buckets for advertisements and telemetry
- Include Alembic migration for schema changes
When multiple receiver nodes are running, the same mesh events (messages,
advertisements) are reported multiple times. This causes duplicate entries
in the Web UI.
Changes:
- Add hash_utils.py with deterministic hash functions for each event type
- Add `dedupe` parameter to messages and advertisements API endpoints (default: True)
- Update dashboard stats to use distinct counts for messages/advertisements
- Deduplicate recent advertisements and channel messages in dashboard
- Add comprehensive tests for hash utilities
Hash strategy:
- Messages: hash of text + pubkey_prefix + channel_idx + sender_timestamp + txt_type
- Advertisements: hash of public_key + name + adv_type + flags + 5-minute time bucket
Members can now have multiple associated nodes, each with a public_key
and node_role (e.g., 'chat', 'repeater'). This replaces the single
public_key field on members with a one-to-many relationship.
Changes:
- Add MemberNode model for member-node associations
- Update Member model to remove public_key, add nodes relationship
- Update Pydantic schemas with MemberNodeCreate/MemberNodeRead
- Update member_import.py to handle nodes list in seed files
- Update API routes to handle nodes in create/update/read operations
- Add Alembic migration to create member_nodes table and migrate data
- Update example seed file with new format
Louis King