* fix(telegram): apply group media policy before inbound downloads
* test(telegram): consolidate inbound media owner regression matrix
* fix(telegram): format media group ingest
* test(telegram): restore media group policy coverage
---------
Co-authored-by: Peter Steinberger <steipete@macos.shared>
Align Telegram artifact reuse and packaged Docker validation with current GitHub Actions and runtime state contracts. Refresh affected regression fixtures and isolate test state.
* test(telegram): cover edited message preview and cache ownership
* fix(telegram): preserve edit previews and refresh message context
* fix(telegram): keep edited messages out of new group history
---------
Co-authored-by: Peter Steinberger <steipete@macos.shared>
* fix(session): preserve reset message payloads
Use the raw command body when it directly matches a reset trigger so code, logs, bracketed text, and line breaks reach the new session intact. Keep structural normalization only for prefixed messages that require it to match.
* fix(session): anchor reset payloads to original input
* fix(session): anchor reset parsing to inbound projections
* fix(session): keep reset result type internal
* test(qa): prove reset payload preservation
---------
Co-authored-by: Vincent Koc <vincentkoc@ieee.org>
Progress-draft lines recovered their detail by string-slicing the label formatToolAggregate had just rendered, and three sites decided "is this a shell tool?" against lowercase spellings only. The Claude CLI sends "Bash", so those checks missed, the slice returned nothing, and every CLI tool call rendered a line with no detail: Telegram printed the icon twice and the tool-summary payload could no longer merge, leaving a second id-less line per call.
formatToolAggregateParts now returns the label with the detail it composed, so a line cannot disagree with its own text, and one predicate owns the shell question across all four sites. Proven live on the Claude CLI backend: two lines with a doubled icon became one.
The group-mentions ack gate returned false whenever a group did not require mentions, conflating group policy with whether this message mentioned the agent. Under the default scope, mentioning the agent in a group configured to answer everything produced no ack at all, so the user had no signal the turn was picked up until the reply arrived.
Behavior change on a default path: those groups now show the 👀 ack (and the lifecycle status reactions when messages.statusReactions.enabled is true) on mentions. Mention-required groups, group-all, unmentioned messages, and off/none are unchanged. The gate no longer reads requireMention, so the parameter is gone from the shared type and all channel call sites.
Proven live on Telegram across three configurations, including a no-mention control confirming group-mentions did not become group-all.
resolveChannelStreamingPreviewToolProgress guessed a mode when streaming.mode
was unset, so it could never be right for every channel: guessing "partial"
dropped Discord and Telegram's explicit progress.toolProgress opt-out, and my
earlier switch to "progress" broke it the other way for Slack, Mattermost, and
Teams, whose real default is partial. With both keys configured and disagreeing,
the wrong one won.
The helper has no channel identity, so it stops guessing and takes the mode the
caller already resolved. All nine call sites pass theirs. Omitting it keeps the
previous configured-mode reading, so untouched callers behave exactly as before.
Reported by ClawSweeper on #116143.
A status headline replaced the rolling tool lines instead of sitting above
them, so a default Discord draft showed one preamble sentence for an entire
tool-heavy turn. Operators reached for `/verbose` to see any activity, which
delivers durable per-tool-call messages and floods the channel.
- Render the headline above the lines; both stay visible in one message.
- Shorten the start gate from 5s to 1.5s. The gate only creates the draft
when the timer fires and finalize cancels it, so quick answers still post
no draft while a 3s tool turn stops being silent.
- Drop Discord's label-gated tool-progress default so
resolveChannelStreamingPreviewToolProgress is the single owner. An explicit
`toolProgress: false` still silences the lines.
- Resolve that toggle against a "progress" mode guess when `streaming.mode`
is unset, so the progress-draft channels stop ignoring an explicit
`progress.toolProgress` opt-out.
Telegram now defaults to `streaming.mode: "progress"` like Discord; set
`"partial"` to keep streamed answer text. Its renderer draws work lines from
the compositor's structured lines, so `rendersRollingLinesNatively` keeps
them out of the composed text rather than printing every line twice.
* fix(telegram): isolate webhook/ingress-worker shutdown errors and guard all cleanup
- Wrap entire shutdown() body in outer try/catch so sync throws never
produce unhandled rejections from fire-and-forget abort listeners.
- bot.stop() rejection is caught and logged; finally block ensures
closeTransportOnce(), noteWebhookStop(), and stopDiagnosticHeartbeat()
always execute regardless of bot.stop() outcome.
- closeTransportOnce() rejection is caught independently in the finally
block so noteWebhookStop() and stopDiagnosticHeartbeat() are not skipped.
- Startup failure path: bot.stop() and closeTransportOnce() get
.catch(() => undefined) so cleanup rejections never mask the original
startup error.
- Both abort-listener call sites use consistent void shutdown() since
the never-reject contract is now internal.
- Ingress worker stop() shares an in-flight stopPromise for idempotency;
worker.terminate() rejections are observed via .catch(() => undefined).
Co-Authored-By: nebulacoder-v8.0 <noreply@zte.com.cn>
* fix(telegram): continue webhook cleanup after sync server.close throw
Extract independently guarded shutdown phases so a synchronous
server.close() failure still runs bot.stop, transport close, status,
and diagnostic heartbeat cleanup. Add sync-throw regression coverage
and a loopback real-behavior proof (negative/positive/valid).
* fix(telegram): clear webhook CI lint, knip, and test-types
Restore curly-safe drain timer clear, stop exporting the unused
shutdown phases interface, and narrow the transport close spy for
tsgo test types.
* chore(telegram): drop committed webhook shutdown proof script
* test(telegram): inline webhook shutdown proof cases without proof script
* fix(telegram): drop stale reply-fence import after main rebase
telegram-reply-fence.ts was deleted on main (f7786a16cf) during the
core drain refactor. The rebased webhook.ts no longer calls these
functions; remove the dangling import so the module resolves.
Co-Authored-By: nebulacoder-v8.0 <noreply@zte.com.cn>
* fix(telegram): continue webhook shutdown after phase failures
Keep each fallible Telegram-owned teardown phase independent so an early failure cannot skip transport, ingress, status, or diagnostic cleanup.
Co-authored-by: Pick-cat <huang.ting3@xydigit.com>
---------
Co-authored-by: nebulacoder-v8.0 <noreply@zte.com.cn>
Co-authored-by: Vincent Koc <vincentkoc@ieee.org>
* fix(telegram): linkPreview:false is ignored on streamed replies
Replies delivered through draft streaming still unfurled the first URL
even with channels.telegram.linkPreview: false. Non-streamed sends
already honored the flag.
createTelegramDraftStream never received linkPreview and never set
link_preview_options, on either its initial sendMessage or its
editMessageText calls. Finalization could not clean it up: it skips the
final edit when the streamed draft text already equals the final text,
which is the common case, so the draft message stays as-is with the
preview attached.
Passes linkPreview from telegramCfg into the draft stream and applies
link_preview_options: { is_disabled: true } on the send and on every
edit, since an edit that omits the field re-enables the preview
server-side. Rich messages keep expressing this as skip_entity_detection
at render time, which already worked.
Call arity is preserved when no preview options apply, so existing edit
assertions are unaffected.
Closes#111525
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* test(telegram): cover streamed reply option parity
Co-authored-by: Zakaria Rahali <zakariarahali288@gmail.com>
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
* fix(telegram): suppress message-tool-only reply fallbacks
Honor message-tool-only delivery when a non-silent Telegram reply is skipped in both ordinary group dispatch and native slash-command turns. Keep the existing visible fallback for genuinely empty native replies and cover both dispatch surfaces with regression tests.
Fixes#90091
* fix(telegram): preserve genuine delivery failure fallbacks
Suppress only non-silent skips owned by message-tool-only delivery. Preserve visible fallbacks when the same turn also encounters a real dispatch or final-delivery failure, with an explicit regression for the mixed failure path.
* fix(telegram): preserve native message-tool delivery errors
Track genuine native slash delivery failures so message-tool-only skip suppression never hides a failed response. Cover the exact mixed empty-skip and real-error path with a regression while preserving true-empty and intentionally cancelled fallbacks.
* test(telegram): preserve synchronous native delivery callbacks
Use the non-mutating dedupe-cache contract when checking unaccepted Telegram updates. Preserve active handler and completed-update deduplication while allowing a real update to be accepted after repeated handler skip probes.
Fixes#105192
Peter Steinberger