The v3 pipeline compacts session history mid-loop (prune_stage +
final-request guard) but only mutates the run's message buffer, never the
session store. Each turn reloads full history and re-compacts from scratch:
message_tokens climb 129k->156k across turns while every turn compacts back
down to ~60k. The lossy compaction differs per run, degrading the agent.
The same missing persistence stalls episodic memory: the cumulative
compaction count never advances, so the episodic worker's idempotency key
(sessionKey:count) is pinned and every cycle after the first is skipped.
Observed on live traffic: 8 run.completed since deploy, 0 new episodic.
Fixes, all reusing existing machinery (no new store methods, no migrations):
- Bug A: emitSessionCompleted reads cumulative GetCompactionCount (matching
the legacy v2 path) instead of the per-run counter that resets to 0.
- Bug B/anti-loop: finalize passes state.Prune.MidLoopCompacted into
maybeSummarize; under pressure it lowers the trigger to a unit-aligned
threshold (compactionInputCap - overhead, same MaxRequestShare the guard
uses) so the compaction is PERSISTED via the existing TruncateHistory +
IncrementCompaction path. Defensive floor prevents over-compaction on
pathological config; tool-result-only bloat still skips (history-only).
- Bug C: SourceID embeds the count (sessionKey:count) so the eventbus dedup
key advances per compaction cycle instead of swallowing rapid same-session
turns within the 5m TTL.
Tests: episodic compaction, maybe_summarize pressure, request budget.
go build (PG + sqliteonly), go vet, go test -race all green.
- Add per-tenant TTS configuration endpoints (GET/POST /v1/tts/config)
with RoleAdmin auth instead of master scope
- Implement TenantTTSResolver for channels to use tenant-specific
TTS providers and auto mode settings
- Add per-agent voice override in channel TTS auto-apply:
- Extend OutboundMessage with AgentID and AgentOtherConfig
- Inject AgentAudioSnapshot in dispatch.go from outbound message
- MaybeApply reads tts_voice_id/tts_model_id from agent context
- Fix events.go to use RunContext.TenantID directly (H3)
- Fix dispatch.go error notification to use sendCtx (H1)
- Update UI to use new /v1/tts/config endpoints
After the first agent run completes, generate a short title from the
user's message using a 10s/50-token LLM call. Title is stored in the
session label column and pushed to clients via session.updated event.
Part A — Channel quota limiter (managed mode):
- DB-backed per-user/group request quotas with in-memory 60s TTL cache
- Config merge priority: Groups > Channels > Providers > Default
- Per-group quota override via channels.telegram.groups[chatID].quota
- Migration 000009: index on channel_requests for quota queries
- Hot-reload quota config via pub/sub (TopicConfigChanged)
Part B — Per-run tool call budget:
- Soft stop at configurable limit (default 25, per-agent override)
- MaxToolCalls field on AgentDefaults + AgentSpec + LoopConfig
- LLM gets one final call to summarize when budget exceeded
Part C — Web UI + config page refactor:
- QuotaSection with provider/channel dropdowns (useProviders, useChannelInstances)
- Config page refactored to vertical sidebar tabs layout
- Categories: General, Quota, Agents, Tools, Connections, Advanced, Raw Editor
- Fixed config.patch RPC to serialize raw JSON + baseHash correctly
- Config change pub/sub broadcast from handleApply/handlePatch
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Multi-agent AI gateway with WebSocket RPC, HTTP API, and messaging channel integrations.
Go port of OpenClaw with multi-tenant PostgreSQL, per-user isolation, security hardening,
and production observability.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Clark Cant