Skip to main content
qa-channel is a repo-local synthetic message transport for automated OpenClaw QA (extensions/qa-channel, private package, excluded from packaged installs). It is not a production channel - it exists to exercise the same channel plugin boundary used by real transports while keeping state deterministic and fully inspectable.

​
What it does

  • Slack-class target grammar:
    • dm:<user>
    • channel:<room>
    • group:<room>
    • thread:<room>/<thread>
  • Shared channel: and group: conversations are surfaced to agents as group/channel room turns, so they exercise the same visible-reply and message-tool routing policy used by Discord, Slack, Telegram, and similar transports.
  • Message sends that omit target and to preserve the inbound conversation kind: direct, group, or channel.
  • Sends in a threaded inbound conversation inherit its thread, including sends to the same explicit root target. topLevel: true or threadId: null suppresses implicit thread and reply inheritance. Explicit thread targets, thread IDs, and replyTo values are preserved.
  • HTTP-backed synthetic bus for inbound message injection, outbound transcript capture, thread creation, reactions, edits, deletes, and search/read actions.
  • Media-bearing streamed replies deliver attachments and the current tool trace together. A later identical text-only final is suppressed only when its caption and tool trace were already delivered successfully.
  • Host-side self-check runner that writes a Markdown report to .artifacts/qa-e2e/.

​
Config

Account keys:
  • enabled - master toggle for this account.
  • name - optional display label.
  • responsePrefix - automatic reply prefix; account overrides win. Accepts a literal, "auto" for the agent identity name, a template such as "[{model}]", or "" to disable an inherited prefix.
  • baseUrl - synthetic bus URL. The account counts as configured once this is set.
  • botUserId - synthetic bot user id used in target grammar (default: openclaw).
  • botDisplayName - display name for outbound messages (default: OpenClaw QA).
  • pollTimeoutMs - long-poll wait window. Integer between 100 and 30000 (default: 1000).
  • mediaMaxMb - per-attachment limit in MiB for inbound bytes and outbound files. Named accounts override the channel root, then agents.defaults.mediaMaxMb supplies the fallback. If none is set, existing media-store and loader defaults apply. An oversized inbound attachment becomes an unavailable-attachment notice for the model. The channel loads each outbound batch before publishing it, so a failed attachment prevents that batch from being sent. Shared reply preparation can first remove oversized local files and retain valid attachments with a warning.
  • allowFrom - sender allowlist (user ids or "*"; default: ["*"]). DMs are always open policy; allowlisted group policy also uses these synthetic sender ids.
  • groupPolicy - shared-room policy: "open" (default), "allowlist", or "disabled".
  • groupAllowFrom - optional shared-room sender allowlist. When omitted under "allowlist", QA Channel falls back to allowFrom.
  • groups.<room>.requireMention - require a bot mention before replying in a specific group/channel room (default: false). groups."*" sets the default; per-room tools / toolsBySender set tool policy overrides.
  • defaultTo - fallback target when none is supplied.
  • actions.messages / actions.reactions / actions.search / actions.threads - per-action tool gating.
Multi-account keys at the top level:
  • accounts - record of named per-account overrides keyed by account id.
  • defaultAccount - preferred account id when multiple are configured.

​
Runners

Host-side self-check (writes a Markdown report under .artifacts/qa-e2e/):
This routes through qa-lab, starts the in-repo QA bus, boots the qa-channel runtime slice, and runs a deterministic self-check. Full repo-backed scenario suite:
The isolated channel-participant-identity-inspection scenario enables execution identity before startup, exercises DM, group, senderless, same- and mixed-participant collect paths, proves an ingress rejection creates no audit rows, and compares JSON plus human CLI inspection across Gateway restart:
Runs scenarios in parallel against the QA gateway lane. See QA overview for scenarios, profiles, and provider modes. Docker-backed QA site (gateway + QA Lab debugger UI in one stack):
Builds the QA site, starts the Docker-backed gateway + QA Lab stack, and prints the QA Lab URL. From there you can pick scenarios, choose the model lane, launch individual runs, and watch results live. The QA Lab debugger is separate from the shipped Control UI bundle.