investigate: OpenClaw as a working option — setup state and a Muse↔OpenClaw comms path that is not laptop SSH

LIST 5b. OpenClaw is live on the laptop (gateway :18789) but its channels are inert (see #1029). Investigate: what OpenClaw can usefully own today, what setup remains, and the comms path between Muse and OpenClaw that does NOT involve SSHing into the laptop — candidates: direct server-to-server API, a Slack channel, or a three-way Muse+Aster+OpenClaw channel. Also note: OpenClaw picks its own name at some point; the design should not hardcode one. Deliverable: setup gap list + comms design recommendation.

Program: Drew's 2026-10-28 directive (full text: hatch-personal memory/topics/id-program-2026-10-28.md).
Working pattern: parallel investigations, then the named synthesis issue merges them into a full design. Route guidance: Workhorse by default; Specialist/Summit for synthesis and ramification-heavy work. A result that says "done" is not the endpoint — the orchestrator reviews it and may commission another round. Deliverables: a markdown design/investigation doc (mermaid/SVG diagrams welcome) linked from this issue, plus follow-up issues for anything buildable. Build a proof of concept where the design calls for one, but do NOT build the permanent feature on top of an unreviewed PoC.

CODE-944

Parent epic: #2055 — fleet expansion & placement program (2026-10-28 directive).

Top priority — ID program wave-1. Drew directed this be picked up in the next runs (2026-10-29): enqueued directly as triage task 3341 in repo-health-code-pilot at queue priority 20 (only breakage-class work at priority 10 outranks it). Wave-1 is the ungated investigations and designs; syntheses and PoCs stay gated on their named inputs. Investigation route: Workhorse.

🤖 Agent note · triaged

Agent: queue worker (repo-worker-code), Lane B shift
Session / ID: queue task 3341
Found via: queue claim
Task: Triage investigate: OpenClaw as a working option — setup state and a Muse↔OpenClaw comms path that is not laptop SSH #2023 — verify the premises (OpenClaw live but channels inert, OpenClaw: connect chat channels (Slack, Telegram, WhatsApp) #1029) and route the investigation.
Verdict: CONFIRMED, with two refinements. Verified live 2026-10-29 ~10:30Z:
- Gateway is live and working locally: openclaw-gateway.service running on the laptop, OpenClaw 2026.10.7, GET /health → {"ok":true,"status":"live"}, 1 agent, 15 sessions, main session active ~10 min before the check — OpenClaw already does local work today; this is not a cold start.
- Channels are not merely inert — none are configured: ~/.openclaw/openclaw.json has an empty channels object. OpenClaw: connect chat channels (Slack, Telegram, WhatsApp) #1029 (Slack/Telegram/WhatsApp) is still OPEN and remains the owning issue for channel setup; nothing to duplicate.
- Refinement 1 — direct API is closer than the issue implies: the gateway already LISTENS on the laptop's tailnet address (100.99.113.103:18789) as well as loopback, and its dashboard URL is the tailnet one. The blockers are on the Muse side: the sandbox is not a tailnet node (its only laptop route is the SSH CONNECT proxy, and an HTTP attempt to the gateway through that proxy was refused — RemoteDisconnected), and gateway auth is a secret-reference token that the CLI path could not resolve (openclaw status degraded read-only; status overview reports loopback auth none). So "direct server-to-server" = fix sandbox tailnet routing (or put a relay on a real tailnet node) + settle gateway token auth — a design question, not a build yet.
- Refinement 2 — Slack is the only candidate both sides can join today: Muse's Slack connector is connected; OpenClaw's Slack channel is unconfigured (that's OpenClaw: connect chat channels (Slack, Telegram, WhatsApp) #1029's first leg). A shared Slack channel also gives the three-way Muse+Aster+OpenClaw variant for free, since Aster can be a channel member rather than a third integration. Any design must stay name-agnostic (per the issue: OpenClaw picks its own name; no hardcoded name in config, channel naming, or prompts).
- Setup gap list (for the design doc): (1) zero channels configured; (2) gateway auth token = secret reference, unresolved outside the service — comms design must name how a remote caller authenticates; (3) sandbox has no non-SSH route to the laptop (proxy is SSH-only); (4) Tailscale exposure shows "off" in openclaw status while the socket binds the tailnet IP — the design must state which is authoritative; (5) what OpenClaw should own is still open — local sessions exist but no inventory of its current work was found in the repo.
Expectation: done — follow-up enqueued (design-doc packet, see below). The investigation deliverable itself (setup gap list expanded + comms design recommendation as a doc/ PR linked from this issue) is the follow-up's scope; do NOT build the permanent comms path on an unreviewed design (issue's own rule).
Links: OpenClaw: connect chat channels (Slack, Telegram, WhatsApp) #1029 (channels), follow-up task enqueued this shift (id in the completion note).

🤖 Agent note · done

Agent: queue worker (repo-worker-code), Lane B shift
Session / ID: queue task 3341
Found via: queue claim
Task: Triage investigate: OpenClaw as a working option — setup state and a Muse↔OpenClaw comms path that is not laptop SSH #2023 complete — verdict CONFIRMED posted above (comment 6494220511): gateway live and working locally, channels unconfigured (empty config, OpenClaw: connect chat channels (Slack, Telegram, WhatsApp) #1029 owns setup), direct-API blockers refined to sandbox tailnet routing + gateway token auth, Slack the only joinable-today path, five-item setup gap list.
Expectation: done — follow-up enqueued as queue task 3381 (design-doc packet for investigate: OpenClaw as a working option — setup state and a Muse↔OpenClaw comms path that is not laptop SSH #2023).

Agent note — parked, merge re-queued (queue task 3381, lane B)

PR #2303 was NOT in the merge queue on fresh check. Re-verified head c4461394 CLEAN/MERGEABLE, checks pass, Codex clean on this head. Re-enqueued — now confirmed in merge queue position 3, state QUEUED.

Resume: when #2303 shows MERGED, complete task 3381 and strip labels on #2023; if ejected again, re-enqueue (stall #1599 / #2403 history).

Agent note — done (queue task 3381, openchamber scheduled lane)

Deliverable: design doc PR #2303 — doc/2026-10-29T12-09-49Z_design_openclaw_comms_path.md (single file, 244 lines), linked from this issue via cross-reference. Covers setup gaps G1–G5 each with an owner, the three-candidate comparison, one recommendation (shared Slack channel now; Aster as a channel member; direct gateway API as phase 2), and name-agnosticism as a hard constraint.
Verified fresh this shift (20:24Z): PR OPEN, MERGEABLE/CLEAN, head c4461394 unchanged, basic: proxy port fallbacks pass, diff matches packet intent.
Not merged — blocker is repo-wide, not this PR: the merge queue ejects entries 2–6 min after enqueue; main has not advanced since ci: run nix ci on merge groups (merge-queue pilot step 1) #2233 merged at 10:57Z, and the queue is empty while green PRs wait. Tracked in #2403 (with ci: Actions check runs stuck in queued across PRs — no runner pickup since ~05:00Z 2026-10-28 #1599 for runner starvation). docs(openclaw): comms-path design — Slack primary, direct API phase 2 (#2023) #2303 was enqueued 4× this shift and ejected each time; it will land when ci: merge queue ejects every PR — no merge since #2233 (10:57Z), entries dropped 2-6 min into AWAITING_CHECKS #2403 is fixed, no further work needed on the branch.
Follow-ups filed by this design: openclaw: enforce token auth for non-loopback gateway callers (comms design G2/G4) #2300 (gateway token auth), fleet: sandbox tailnet route to the laptop for the direct-API comms path (comms design G3) #2301 (sandbox tailnet route), openclaw: inventory the existing sessions and what the agent already owns (comms design G5) #2302 (session inventory). Channel setup stays OpenClaw: connect chat channels (Slack, Telegram, WhatsApp) #1029's.
Queue task: completed pr_opened (issue left open for the orchestrator review the program description asks for).

🤖 Agent note · outcome recorded (for issue #2301 / gap G3)

Agent: queue worker (repo-worker-code), lane code-k · task 7888
Recording the G3 outcome on this parent issue per issue fleet: sandbox tailnet route to the laptop for the direct-API comms path (comms design G3) #2301's acceptance criteria: the sandbox → laptop tailnet route is now demonstrated. The sandbox is enrolled in the tailnet via Muse's Tailscale connector (100.115.16.3, outbound TCP via the runtime tunnel), the laptop amd-6 is at 100.99.113.103, the OpenClaw gateway listens on 100.99.113.103:18789, and an HTTP request from the sandbox to that address returned 200 (verified 2026-11-03). Full evidence is on issue fleet: sandbox tailnet route to the laptop for the direct-API comms path (comms design G3) #2301, which is closed on this basis.

For the record (no state change — issue #2023 stays closed): the OpenClaw session/ownership inventory requested by its child issue #2302 is published at doc/2026-11-03T23-32-00Z_inventory_openclaw_sessions_ownership.md (PR #5206, merged 6aeb038; queue task 8905).

Navigation Menu

Repository navigation

investigate: OpenClaw as a working option — setup state and a Muse↔OpenClaw comms path that is not laptop SSH #2023

82 remaining items

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

investigate: OpenClaw as a working option — setup state and a Muse↔OpenClaw comms path that is not laptop SSH #2023

Description

Activity

developing-today commented on Oct 29, 2026

dezren39 commented on Oct 29, 2026

dezren39 commented on Oct 29, 2026

dezren39 commented on Oct 29, 2026

dezren39 commented on Oct 29, 2026

82 remaining items

dezren39 commented on Oct 29, 2026

dezren39 commented on Oct 29, 2026

dezren39 commented on Nov 3, 2026

dezren39 commented on Nov 3, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions