The opencode shell wrapper sets HTTPS_PROXY for z.ai API egress, and
child processes inherit it. The proxy rejects local IPs with 403,
breaking Proxmox API calls from the backend. Explicitly clear all
proxy env vars so the backend always talks direct to local infra.
Manifest-backed MCP tools, prompts, and resources with surface affordance contracts; agent capability manifest and governance projection; API contract tests and capability route projection; operations-loop and intelligence-funnel telemetry; release-control subsystem documentation, registry, and tooling; licensing and configuration.
The liveness handler called SchedulerHealth(), which refreshes provider
instance caches under the monitor's write lock and snapshots queues and
breakers on every probe — it only needed the dead-letter count. Add
Monitor.DeadLetterCount() (queue-local mutex only) and use it.
The hot-dev watchdog killed the backend after 2 missed 3s probes; on a
dev box running race-detector test suites and builds that is routine
CPU starvation, and each kill destroyed any in-flight Assistant chat
stream (observed repeatedly during the chat-feel audit). Default the
miss threshold to 4 (~20s) — genuine wedges still die quickly.
Mock mode seeded rich in-memory chart history but passed a nil store to the
seeder and the live tick, so the sqlite metrics store held almost nothing:
vmware/truenas/docker resources got one row per poll from the platform
ingestion paths and mock Proxmox guests got nothing at all (the unified sync
skips every resource in mock mode). Performance reports rendered
"Data Points: 0" for PVE guests and "Samples: 1" elsewhere, with no charts.
Re-enable store seeding behind an explicit PULSE_MOCK_SEED_METRICS_STORE
opt-in that scripts/hot-dev.sh and scripts/toggle-mock.sh export exactly
where they point PULSE_DATA_DIR at the isolated tmp/mock-data dir. Without
the opt-in (a production install flipping PULSE_MOCK_MODE on its real data
dir) the store stays untouched, which is what the old nil guard protected.
Replace the dormant dense seeding policy (every in-memory timestamp written
to both hourly and daily tiers, ~11M rows at current fixture scale) with
tier-correct backfill driven by the deterministic mock.SampleMetric runtime:
daily 30d at 4h spacing, hourly 7d at 2h, minute 24h at 15m, raw left to the
live tick (now also store-connected under the opt-in). Timestamps sit on the
spacing grid and a new Store.MaxTimestampsForTier coverage query fills only
the gap since the previous boot, so restarts neither duplicate rows nor
re-pay the seed: a fresh seed wrote 264,960 rows in 8.4s (71MB) and a
restart wrote only the gap plus fixture series whose mock IDs are not
boot-stable (k8s pod names, ceph FSID, a pre-existing generator defect).
Verified live in mock mode: 30-day PDF reports for a mock PVE VM
(checkout-web-01), a vSphere VM, the TrueNAS host, and a Docker container
now show 1260/1260/1260/540 data points with rendered time-series charts.
Mock and real mode shared tmp/dev-config, so toggling between them left
residue each way: real connections (and a live local agent) showed in
Settings during mock, and mock-cluster alerts lingered in real mode.
The toggle script had an isolation flag but it only reached the
standalone runtime, never managed-hot-dev.
hot-dev.sh now routes mock mode to tmp/mock-data and real mode to
tmp/dev-config, choosing the dir from the canonical PULSE_MOCK_MODE flag
that toggle-mock writes to tmp/dev-config/.env. The flag is read from
that file authoritatively rather than from the environment, because a
stale PULSE_MOCK_MODE exported into the hot-dev supervisor from a prior
mock run would otherwise keep real mode pinned to the mock dir.
toggle-mock status now reports the isolated mock dir for hot-dev
runtimes so it matches what the backend actually uses.
Verified a full round trip: real -> dev-config (real connections only),
mock -> mock-data (mock data only, no real bleed), real -> dev-config
(no mock-cluster residue). Trade-off: real metrics history no longer
accrues during mock UI sessions, which is the right call for clean
dev switching.
Keep the managed dev runtime on loopback unless PULSE_DEV_LAN=true is set, so installed LAN agents cannot accidentally attach to a developer session. Harden managed status checks for restricted local shells. Refs architecture post-RC canonicalization follow-up.
Fix the local hot-dev backend monitor so a missing Pulse process is counted safely under pipefail, keep backend launch stderr in the debug log, and govern the new runtime helper with focused smoke coverage.
The supervisor's backend health monitor in scripts/hot-dev.sh only
checked whether a `./pulse` process existed via pgrep. If the Go binary
hung, panicked into recovery without exiting, or failed to bind :7655
while the process stayed alive, the monitor saw a healthy process count
and never restarted, leaving the dev frontend talking to a dead listener.
Hit that twice in one session.
Add an HTTP probe of /api/health on the dev port. After two consecutive
5s polls where the process exists but /api/health is unreachable, kill
the process and restart the same way the dead/duplicate branches do.
Cut the poll interval from 10s to 5s so the worst-case unresponsive
window is ~10s instead of ~30s. Bring deployment-installability into
line by naming alive-but-unresponsive recovery as part of the dev-runtime
orchestration boundary, and cover the new branch with static-source
assertions in scripts/tests/test-hot-dev-bg.sh.
When scripts/watch-backup.sh exists (gitignored, local only), hot-dev
automatically starts a file watcher that backs up source files on every
save to ~/.pulse-backups/. Protects against accidental data loss from
git checkout or other mishaps, regardless of what tool makes the change.
- hot-dev.sh: Fix hostname -I for macOS, use ifconfig instead
- hot-dev.sh: Fix PULSE_AUDIT_DIR for mock mode
- hot-dev.sh: Use PULSE_REPOS_DIR for Pro module detection
- dev-check.sh: Fix pgrep -c (not supported on macOS)
- dev-check.sh: Use /tmp/pulse-debug.log on macOS instead of journalctl
- Update internal/api docs to use env var paths
The agent was crashing with 'fatal error: concurrent map writes' when
handleCheckUpdatesCommand spawned a goroutine that called collectOnce
concurrently with the main collection loop. Both code paths access
a.prevContainerCPU without synchronization.
Added a.cpuMu mutex to protect all accesses to prevContainerCPU in:
- pruneStaleCPUSamples()
- collectContainer() delete operation
- calculateContainerCPUPercent()
Related to #1063
Add ability for users to describe what kind of agent profile they need
in natural language, and have AI generate a suggestion with name,
description, config values, and rationale.
- Add ProfileSuggestionHandler with schema-aware prompting
- Add SuggestProfileModal component with example prompts
- Update AgentProfilesPanel with suggest button and description field
- Streamline ValidConfigKeys to only agent-supported settings
- Update profile validation tests for simplified schema
Update the "Kiosk / Dashboard" token preset description to mention
the ?kiosk=1 URL parameter that hides navigation and filter panels,
making it clearer how to set up a wall display.
Addresses feedback from #1102
Allow users to set custom disk usage thresholds per mounted filesystem
on host agents, rather than applying a single threshold to all volumes.
This addresses NAS/NVR use cases where some volumes (e.g., NVR storage)
intentionally run at 99% while others need strict monitoring.
Backend:
- Check for disk-specific overrides before using HostDefaults.Disk
- Override key format: host:<hostId>/disk:<mountpoint>
- Support both custom thresholds and disable per-disk
Frontend:
- Add 'hostDisk' resource type
- Add "Host Disks" collapsible section in Thresholds → Hosts tab
- Group disks by host for easier navigation
Closes#1103
Major changes:
- Add audit_logging, advanced_sso, advanced_reporting features to Pro tier
- Persist session username for RBAC authorization after restart
- Add hot-dev auto-detection for pulse-pro binary (enables SQLite audit logging)
Frontend improvements:
- Replace isEnterprise() with hasFeature() for granular feature gating
- Update AuditLogPanel, OIDCPanel, RolesPanel, UserAssignmentsPanel, AISettings
- Update AuditWebhookPanel to use hasFeature('audit_logging')
Backend changes:
- Session store now persists and restores username field
- Update CreateSession/CreateOIDCSession to accept username parameter
- GetSessionUsername falls back to persisted username after restart
Testing:
- Update license_test.go to reflect Pro tier feature changes
- Update session tests for new username parameter
- Login.tsx: Use apiClient.fetch with skipAuth to avoid auth loops
- router.go: Skip CSRF validation for /api/login endpoint
- hot-dev.sh: Detect encrypted files before generating new key to prevent data loss
- crypto.go: Add runtime validation to Encrypt() that verifies the key file
still exists on disk before encrypting. If the key was deleted while Pulse
is running, encryption now fails with a clear error instead of creating
orphaned data that can never be decrypted.
- hot-dev.sh: Auto-generate encryption key for production data directory
(/etc/pulse) when HOT_DEV_USE_PROD_DATA=true and key is missing. This
prevents startup failures and ensures encrypted data can be created.
- Added test TestEncryptRefusesAfterKeyDeleted to verify the protection works.
- Add host metadata API for custom URL editing on hosts page
- Enhance AI routing with unified resource provider lookup
- Add encryption key watcher script for debugging key issues
- Improve AI service with better command timeout handling
- Update dev environment workflow with key monitoring docs
- Fix resource store deduplication logic
- Add AI service with Anthropic, OpenAI, and Ollama providers
- Add AI chat UI component with streaming responses
- Add AI settings page for configuration
- Add agent exec framework for command execution
- Add API endpoints for AI chat and configuration
