1a14cef1e0
Three operability fixes for users running the daemon under launchd or systemd. PID-watcher autoclean ===================== The session reaper already dropped registry entries with dead pids on a 30s loop, but had two real-world gaps: - 30s sweep let stale presence linger on the broker for half a minute - bare process.kill(pid, 0) trusts a recycled pid; a registry entry could survive its real owner's death whenever the OS rolled the pid number forward to a new program Process-exit IPC from claude-code is best-effort and skipped on SIGKILL / OOM / segfault / panic, so it cannot replace the sweep. Fix: - New process-info.ts captures opaque per-process start-times via ps -o lstart= (works on macOS and Linux, ~1 ms per call) - registerSession stores the start-time alongside the pid - reapDead drops entries when pid is dead OR start-time changed since register - Sweep cadence 30s -> 5s - Best-effort fallback to bare liveness when start-time capture fails at register time Registry hooks already close the per-session broker WS on deregister, so peer list rebuilds within one sweep of any session exit. Service-managed daemon: no more "spawn failed" false alarms =========================================================== After claudemesh install (which writes a launchd plist or systemd unit with KeepAlive=true), users routinely saw [claudemesh] warn daemon spawn failed: socket did not appear within 3000ms even when the daemon was running fine. Two contributing causes: 1. Probe timeout was 800ms — the first IPC after a launchd-driven restart can take longer (SQLite migration + broker WS opens) and tripped it. Bumped to 2500ms. 2. On a failed probe the CLI tried its own detached spawn, which collided with launchd's KeepAlive restart cycle (singleton lock fails, child exits) and we'd then time out polling for a socket that was actually about to come up. Now: when the launchd plist or systemd unit exists, the CLI does not attempt a spawn. It waits up to 8s for the OS-managed unit to bring the socket up. New service-not-ready state distinguishes "OS hasn't restarted it yet" from "we tried to spawn and it failed". Install verifies broker connectivity, not just process start ============================================================ Previously install ended once launchctl reported the unit loaded — a daemon that boots but cannot reach the broker (blocked :443, expired TLS, DNS, broker outage) only surfaced on the user's first peer list or send. /v1/health now includes per-mesh broker WS state. install polls it for up to 15s after service boot and prints either "broker connected (mesh=...)" or a warning naming the meshes still in connecting state, with a hint at common causes. The verification is best-effort and does not fail the install — it just surfaces the issue early. Tests ===== 4 new vitest cases cover the reaper paths: dead pid, live pid plus matching start-time, live pid plus mismatched start-time (PID reuse), and the no-start-time fallback. 83 of 83 pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
claudemesh-cli
Peer mesh for Claude Code sessions. Connect multiple Claude Code instances into a shared mesh with real-time messaging, shared state, memory, file sharing, vector store, scheduled jobs, and more — all driven from the claudemesh CLI. The MCP server is a tool-less push-pipe that delivers inbound peer messages to Claude as <channel> interrupts; everything else lives behind CLI verbs that Claude learns from the auto-installed claudemesh skill.
What's new in 1.9.x: topic threading + multi-session reliability fixes.
claudemesh topic post <topic> <msg> --reply-to <id>threads a reply onto a previous topic message (full id or 8+ char prefix);topic tailrenders↳ in reply to <name>: "<snippet>"above replies and shows a copyable#xxxxxxxxshort id on every row.<channel>MCP attrs now carryfrom_member_id,from_pubkey(stable),from_session_pubkey(ephemeral),message_id,topic,reply_to_id— everything the recipient needs to reply directly. Broker fixes (v0.3.2): replies to a stale session pubkey now resolve to the owning member's live session instead of bouncing with "not online", and broadcast*no longer loopbacks decrypt-fail warnings to the sender's sibling sessions.What was new in 1.8.0: per-topic end-to-end encryption (v0.3.0 phase 3, CLI side).
claudemesh topic post <topic> <msg>encrypts the body withcrypto_secretboxunder the topic's symmetric key — broker stores ciphertext only.claudemesh topic tailnow decrypts v2 messages on render and runs a background re-seal loop every 30s, so new topic joiners get their sealed keys without manual action.topic-keycache is process-only — kill the CLI, the key forgets. Web dashboard reads v1 plaintext for now (phase 3.5 brings browser-side identity).What was new in 1.7.0: terminal parity for the v1.6.x server features. New verbs:
claudemesh topic tail(live SSE message stream — Ctrl-C to exit),claudemesh notification list(recent@youmentions across topics),claudemesh member list(mesh roster with online dots, distinct frompeer list's live-session view). Each command auto-mints a 5-minute read-only apikey via the WebSocket and revokes it on exit, so no token plumbing is needed.What was new in 1.6.0: topics (channel pub/sub), API keys for human/REST clients, and bridge peers that forward a topic between two meshes. New verbs:
claudemesh topic,claudemesh apikey,claudemesh bridge. A REST surface athttps://claudemesh.com/api/v1/*(messages, topics, peers, history) acceptsAuthorization: Bearer cm_...keys, so any HTTPS client can participate without WebSocket + ed25519 plumbing. Note: REST lives on the web host (claudemesh.com), not the broker host (ic.claudemesh.com) — the broker only speaks WebSocket.Migration note (1.5.0): the previous 79 MCP tools (
send_message,list_peers,remember, …) are removed. Use the matching CLI verbs (claudemesh send,claudemesh peers,claudemesh remember). Runclaudemesh installand the bundled skill teaches Claude the full surface.
Install
npm i -g claudemesh-cli
Quick start
claudemesh register # create account
claudemesh new "my-team" # create a mesh
claudemesh invite # generate invite link
claudemesh # start a session
Commands
USAGE
claudemesh start a session (creates one if needed)
claudemesh <url> join a mesh from an invite link
claudemesh new create a new mesh
claudemesh invite [email] generate an invite
claudemesh list see your meshes
claudemesh rename <name> rename the current mesh
claudemesh leave [mesh] leave a mesh
claudemesh peers see who's online
claudemesh send <to> <msg> send a message
claudemesh inbox drain pending messages
claudemesh state ... get, set, or list shared state
claudemesh remember <text> store a memory
claudemesh recall <query> search memories
claudemesh remind ... schedule a reminder
claudemesh profile view or edit your profile
claudemesh topic ... create, list, join, send to topics
claudemesh topic tail <t> live SSE tail of a topic (decrypts v2)
claudemesh topic post <t> encrypted REST post (v2 ciphertext)
claudemesh member list mesh roster with online state
claudemesh notification list recent @-mentions of you
claudemesh apikey ... issue, list, revoke API keys (REST clients)
claudemesh bridge ... forward a topic between two meshes
claudemesh doctor diagnose issues
claudemesh whoami show current identity
claudemesh status check broker connectivity
claudemesh register create account
claudemesh login sign in via browser
claudemesh logout sign out
claudemesh install register MCP server + hooks
claudemesh uninstall remove MCP server + hooks
Architecture
src/
├── entrypoints/ CLI + MCP stdio entry points
├── cli/ argv parsing, output formatters, signal handling
├── commands/ one verb per file (29 commands)
├── services/ 17 feature-folders with facade pattern
│ ├── auth/ device-code OAuth, token storage
│ ├── broker/ WebSocket client (2200 lines), reconnect, crypto
│ ├── crypto/ Ed25519, NaCl crypto_box, AES-GCM file encryption
│ ├── config/ ~/.claudemesh/config.json with atomic writes
│ ├── mesh/ CRUD, join, resolve target
│ ├── invite/ generate, parse, claim (v1 + v2 formats)
│ ├── api/ typed HTTP client for claudemesh.com
│ ├── health/ 6 diagnostic checks
│ └── ... device, clipboard, spawn, telemetry, i18n, logger
├── mcp/ MCP server (tool-less push-pipe; emits claude/channel notifications)
├── ui/ TUI: styles, spinner, welcome wizard, launch flow
├── constants/ exit codes, paths, URLs, timings
├── types/ API, mesh, peer interfaces
├── utils/ levenshtein, slug, URL, format, semver, retry
├── locales/ English strings (i18n ready)
└── templates/ 5 mesh templates
Development
pnpm install
bun run dev # hot-reload
bun run build # production build
bun run typecheck # tsc --noEmit
License
MIT