ACP Session Init Failed

[topaz crescent]

ACP error (ACP_SESSION_INIT_FAILED): Session is not ACP-enabled: agent:main:discord:channel:xxxxxxxxxx
next: If this session is stale, recreate it with /acp spawn and rebind the thread.

[stone ledge]

That usually means this Discord thread is pointing at a normal session key, not an ACP session.

In this same thread, try:

1. /acp spawn codex --bind here
   or replace codex with your ACP agent
2. If it was previously bound and still acts weird, run /unfocus first, then spawn again
3. If you are spawning from a parent channel into a new thread, make sure channels.discord.threadBindings.spawnAcpSessions=true

Also, what does openclaw --version show?

If respawning still fails, paste:

• openclaw status --all
• /acp doctor

Docs:

• https://docs.openclaw.ai/tools/acp-agents

• https://docs.openclaw.ai/channels/discord

[topaz crescent]

⚠️ Session binding adapter failed to bind target conversation

[topaz crescent]

Tried spawn bind, got this error:
⚠️ Session binding adapter failed to bind target conversation

[stone ledge]

Ah, that helps. ACP spawn is getting far enough to try the bind, and the binding step is what's failing.

Can you paste these next:

1. openclaw --version
2. /acp doctor
3. openclaw status --all

And one quick isolation test:

4. /acp spawn codex --thread off

If --thread off works, ACP itself is fine and the problem is specifically Discord conversation/thread binding.

Then check these config bits:

• channels.discord.threadBindings.enabled: true

• if you want thread-style ACP binds, also channels.discord.threadBindings.spawnAcpSessions: true

Also worth trying, from inside this thread:

• /acp spawn codex --thread here

If that still fails on a current version, I'm honestly suspicious this is a bug in the Discord binding path. Docs:

• https://docs.openclaw.ai/tools/acp-agents
• https://docs.openclaw.ai/channels/discord
• https://docs.openclaw.ai/help/faq

[topaz crescent]

Version 4.12

Doctor result:
ACP doctor:

configuredBackend: acpx
activeRuntimeSessions: 0
runtimeIdleTtlMs: 7200000
evictedIdleRuntimes: 0
activeTurns: 0
queueDepth: 0
turnLatencyMs: avg=0, max=0
turnCounts: completed=0, failed=0
errorCodes: (none)
registeredBackend: acpx
runtimeDoctor: ok (embedded ACP runtime ready)
runtimeDoctorDetail: agent=codex
runtimeDoctorDetail: command=npx @zed-industries/codex-acp@^0.11.1
runtimeDoctorDetail: cwd=/Users/xxxx/.openclaw/workspace
runtimeDoctorDetail: protocolVersion=1
healthy: yes
capabilities: session/set_config_option, session/set_mode, session/status

[stone ledge]

Nice, that narrows it down a lot.

Your ACP backend looks healthy, so this does not look like a Codex/acpx problem. It looks like the Discord binding layer.

Two quick notes:

• Version 4.12 sounds pretty old, if that is really the full openclaw --version output. If so, I would update first.
• On a healthy/current build, /acp spawn codex --bind here in a Discord thread should work.

Can you run these next:

1. paste the exact full output of openclaw --version
2. /acp spawn codex --thread off
3. /acp spawn codex --thread here
4. openclaw status --all

Also paste your redacted JSON5 for:

channels: {
  discord: {
    threadBindings: {
      enabled: true,
      spawnAcpSessions: true,
    },
  },
}

If --thread off works, but --bind here and --thread here both fail, I’d treat that as a likely Discord binding bug.

If it comes to that, please file it here with the exact version, doctor output, commands tried, and error text:
https://github.com/openclaw/openclaw/issues

[topaz crescent]

Found the issue, I made the bindings.accountId wrong

[stone ledge]

Nice, that’ll do it. An accountId mismatch in bindings[] can absolutely make the bind path fall over like this.

Glad you caught it 🎉

If this fixed it, please right-click the solution message → Apps → ✅ Mark Solution so Answer Overflow can index it for the next person.

[topaz crescent]

@unborn galleon Are you broken?