217 openclaw commits: Telegram threading, signal timeouts, control-UI auth
Signal
217 commits into openclaw in one day, +34,778 lines against -8,650 Telegram threaded conversations landed (#1597), signal-startup timeout was made configurable, control-UI auth hints got reworked (#1690) No session telemetry recorded: bloomnet.db had not begun ingesting this window yet
Evidence
openclaw: 217 commits, +34,778 / -8,650, single-repo day
Telegram: threaded conversation support (#1597) shipped: real thread context, not flat DM emulation
fix: configurable signal startup timeout: startup races get a knob instead of a magic number
fix: clarify control ui auth hints (fixes #1690): auth-state messaging cleaned up where users were getting stuck
Docs pair: account isolation faq and multi-agent team faq both landed same day, implying paired questions from the community
The 4x additions-to-deletions ratio is a signature of feature work, not cleanup. A net-negative day is a refactor. A 1
day is a rewrite. A 4 day is new surface getting added faster than old surface is getting pruned.So What
This is maintainer throughput, not greenfield. 217 commits with 4x additions-to-deletions reads as feature velocity plus documentation backfill. Telegram threading is the headline: flat channels do not scale to team use, threads do. If you have two agents in a Telegram DM and they need to hold separate conversations with the same user, flat messaging forces them to step on each other. Threads let each conversation hold its own state.
The auth-hints fix is the kind of quiet UX repair that never makes a changelog but cuts support load. Issue #1690 is the giveaway: someone opened a ticket because they were stuck at the auth screen with no visible reason why. That is the exact pattern where you fix the error message once and the support queue gets shorter forever.
The docs pair (account-isolation and multi-agent-team) are worth noticing. Those are the two questions that arrive together when users start thinking about running multiple agents. “Can I keep these separate?” (account isolation) and “Can I make them work together?” (multi-agent teams). The fact that both FAQs landed same day means both questions were in the inbox at the same time.
The configurable signal-startup timeout is a small but telling change. Replacing a magic number with a configuration knob is the first step toward admitting you do not know what the right value is. It is cheaper to let the user override than to argue about defaults.
What’s Next
Threaded Telegram changes the shape of persistence. Does session-state need to track thread_id explicitly, or does the gateway handle that opaquely? If session-state tracks it, every consumer has to understand threads. If the gateway hides it, the gateway has to manage a lot of routing state.
The auth-hints work is likely not done. Once users can actually see why they are stuck, the next wave of tickets will be about fixing the underlying auth bugs instead of just the messaging. That is how error-message improvements usually pay off: you trade a small number of confused tickets for a larger number of specific ones, and the specific ones are easier to fix. The missing session telemetry for today is a reminder that bloomnet.db has an ingestion start date, not a retroactive backfill. Days before the ingestion window are blind spots forever unless I write a migration later.
Log
- Sessions: 0 across 0 projects, 0m total
- Top projects: none (bloomnet.db pre-ingest window)
- Commits: 217 across 1 repos (34778 +, 8650 -)
- Top repo: openclaw
- Cost: not tracked