Setup
Connect Your Agent to Google Chat
A practical setup + troubleshooting guide for adding your heyron agent to Google Chat spaces without getting stuck.
Community pattern: teams working in Google Workspace want to keep collaboration in Google Chat, but setup gets blocked by app visibility, permissions, or wrong space settings. This guide gives you a clean path.
Reality check: Google Chat integration depends on your Google Workspace app permissions and who can install apps to spaces. If your org locks this down, ask your admin first.
Fast setup flow (5 steps)
1) Confirm your target conversation type
Decide whether you need the agent in a DM, a group chat, or a Space. Many failures come from testing in one type while configuring for another.
2) Install/add the bot app in Google Chat
In Google Chat, click + New chat or open your Space settings, then add the heyron/OpenClaw bot integration your team is using.
3) Verify app permissions in Workspace
If the app is missing, your Workspace policy may block app installs. Ask your admin to allow the app for your OU/group.
4) Run a tiny canary test
Send one clean test message in the exact destination you care about:
@YourAgent reply with exactly: "GOOGLE_CHAT_CANARY_OK"
If this fails, do not continue with bigger workflows yet.
5) Lock behavior expectations
Once canary passes, send a short operating prompt (tone, language, response limits, and task scope) so your shared space stays clean and consistent.
Most common failure modes
- "I can’t find the app": app blocked by Workspace admin policy.
- "Bot was added but silent": tested in a different space/thread than where app was added.
- "Works for me, not my partner": permission is user- or group-scoped, not org-wide.
- "Replies are inconsistent": separate conversations have separate context, like dashboard vs chat channels elsewhere.
Known-good bootstrap prompt for shared spaces
For this Google Chat space:
1) Reply in concise bullet points.
2) Ask at most ONE clarification question before acting.
3) If blocked by permissions/integration limits, say exactly what is blocked and the fastest human workaround.
4) End each response with: "Next best step: ..."
Important: if you want both you and your partner to use the same agent reliably, test both accounts in the same space with the same canary prompt before real work.
Advanced edge cases (when basic checks pass but it still fails)
- App only responds when directly mentioned: some spaces are configured so apps do not answer ambient messages. Use explicit
@YourAgent mentions in your operating pattern.
- Thread mismatch: canary was run in main space, real task posted in a thread (or vice versa). Treat each thread as its own context lane.
- External/guest membership limits: mixed-domain spaces can have tighter app policies than internal spaces.
- One user works, another fails: OU/group rollout lag or app access policy not fully propagated yet. Re-test after policy sync, then escalate with account-level evidence.
- Notifications make it look silent: app replied in-thread but users were watching main timeline only. Confirm by opening the exact thread and sorting by newest.
Partner collaboration checklist (2 humans + 1 agent)
Recent community pattern: one person can use the agent, but their partner cannot in the same workspace. Run this exact checklist before deeper debugging:
- Use one shared Space (not separate DMs) for your first real test.
- Have both humans run the same canary in that exact Space/thread:
@YourAgent reply with exactly: "GOOGLE_CHAT_CANARY_OK".
- Compare behavior side-by-side: if one succeeds and one fails, this is usually account/OU policy scope, not an agent failure.
- Verify both users can see the app card in Space members/integrations.
- Lock one working pattern (for example: always @mention in thread) and document it in the space topic.
Common trap: testing in your own DM and your partner testing in a Space can look like "different agent behavior." It's usually different context + policy scope.
When to escalate in #help
If setup still fails, share:
- Whether you tested DM, group chat, or Space (and thread vs main timeline)
- Whether the app appears in Google Chat search for each affected user
- The exact canary prompt and exact response/error (copy/paste, no paraphrase)
- Whether direct @mention works while plain messages fail
- Whether one partner succeeds while the other fails in the same space/thread
- Whether your Workspace admin confirmed app access policy + OU/group scope
Bottom line: most Google Chat failures are policy + destination mismatch problems, not model problems. Do the canary in the exact final destination (including thread context), then scale.