| MSG-54 |
llmmsg-srv |
KPI Track A: ADD pg views (usage.v_kpi_tokens_7d + context.v_kpi_compact_7d) + extend scrp-kpiweekly-whey (email + snapshot/live-read) |
backlog |
high |
bin-whey-cc |
2026-06-16 01:23 |
| MSG-38 |
llmmsg-srv |
POST /unregister route: deregister permanence for VENUSINF waves - {agent,caller,[bearer]}; txn validate->capture-PM-AROs->deleteAroByAgent->electPm-each->deleteRosterRow; PRESERVE cursors (optional purge_cursor:true default false); idempotent 200 already_absent; auth self-dereg ungated + admin-dereg edge-bearer interim soft-dep #599; HTTP-only (no MCP tool). Replaces hand-SQL roster DELETE dereg-at-cutover. |
backlog |
high |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| MSG-32 |
llmmsg-srv |
init-db.sh drift: stale messages schema (generated tag + phantom cols) contradicts hub.mjs self-create / live DB - reconcile or demote |
backlog |
high |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| MSG-13 |
llmmsg-srv |
Hub leaks direct DMs into sender's home ARO: kind=dm fanned to origin_aro roster (cc-context-monitor-venus DMs reaching nw-venus-cc) |
backlog |
high |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| MSG-5 |
llmmsg-srv |
Post-/compact re-register reliability: hook-fire blind spot + hook-independent shim re-register net |
backlog |
high |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #603 |
llmmsg-srv |
Alert when a CC session is blocked on interactive prompt (AskUserQuestion menu / permission approval / ExitPlanMode): wire CC Notification hook (or transcript-watcher fallback) to DM elazar via llmmsg-srv naming the blocked agent+host. Solves operator having to hunt tmux windows. Research dispatched to claude-code-guide. |
backlog |
high |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #599 |
llmmsg-srv |
Close aro_config auth gap: token-gate POST /aro_config + other privileged hub writes so a forged caller=elazar fails (coder-applog-cc exploited it to flip aro:applog PM). Elazar approved ship. |
backlog |
high |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| MSG-35 |
llmmsg-srv |
Spurious unread-poke loop: read_unread cursor-ack is fire-and-forget (race) + drain-on-other-tools acks only on next poll + poker lacks debounce/idle-gate |
backlog |
high |
pm-llmmsgsrv-cc |
2026-06-14 04:58 |
| MSG-12 |
llmmsg-srv |
Make context-budget nag idempotent: stop re-nagging every turn + stop stacking /compact crons after compact |
backlog |
high |
pm-llmmsgsrv-cc |
2026-06-06 06:53 |
| MSG-10 |
llmmsg-srv |
Codex receive-bridge: sidecar tmux-poke so Codex agents auto-drain llmmsg unread (no live push) |
backlog |
high |
pm-llmmsgsrv-cc |
2026-06-06 06:24 |
| MSG-11 |
llmmsg-srv |
└─ HARD PREREQ: derive + test a Codex-specific ready-prompt regex for idle-gating (CC matcher will NOT transfer; wrong matcher = corrupted-buffer poke). No send-keys goes live until validated. |
backlog |
high |
nw-whey-cc |
2026-06-06 06:24 |
| #572 |
llmmsg-srv |
ca.sh codex launcher: run codex on llmmsg-srv (ca= identity, sandbox-off, pull-only inbound, push-to-ccs) - fleet |
inProgress |
high |
pm-llmmsgsrv-cc |
2026-06-05 06:51 |
| #573 |
llmmsg-srv |
└─ Design + LIVE-TEST ca.sh: reads ca= from .agent-name, codex --dangerously-bypass-approvals-and-sandbox, injects -c mcp_servers.llmmsg-srv.env.LLMMSG_AGENT_NAME + LLMMSG_HUB_URL; PROVE shim auto-registers under ca name (roster check) before calling done |
backlog |
high |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #574 |
llmmsg-srv |
└─ Land reviewed ca.sh in sh.git (after PM draft review); commit+push |
backlog |
high |
bin-whey-cc |
2026-05-29 02:58 |
| #577 |
llmmsg-srv |
└─ Fix static ~/.codex/config.toml footgun: wrong key LLMMSG_AGENT->LLMMSG_AGENT_NAME + LLMMSG_SRV_HOST 127.0.0.1->public; bare codex (not via ca.sh) currently falls back to cc= in /opt/llmmsg-srv = registers as pm-llmmsgsrv-cc on loopback. Non-blocking; ca.sh -c overrides already bypass it |
backlog |
normal |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #576 |
llmmsg-srv |
└─ lezama ca.sh deploy: GCABA box - get Elazar explicit OK before nw-lezama-cc syncs; codex AGENTS.md convention (aro_join on start, cm/cms -> read_unread) |
deferred |
normal |
pm-llmmsgsrv-cc |
2026-06-04 15:56 |
| #575 |
llmmsg-srv |
└─ Deploy/verify ca.sh on venus (sh.git pull); confirm a codex test session registers + hub URL resolves remote |
backlog |
normal |
nw-venus-cc |
2026-05-29 02:58 |
| #589 |
llmmsg-srv |
lezama split-brain: island-b hub (llmmsg-srv.service) auto-started, isolated lezama agents from whey hub |
backlog |
high |
pm-llmmsgsrv-cc |
2026-05-29 17:45 |
| #565 |
llmmsg-srv |
Memory-discipline policy: make enforcement discoverable fleet-wide (whey/venus/lezama) |
backlog |
high |
pm-llmmsgsrv-cc |
2026-05-29 01:56 |
| #569 |
llmmsg-srv |
└─ Per-host global CLAUDE.md: add memory-discipline.md to Lazy-loaded + 2-line enforcement pointer (whey/venus/lezama, one commit each) |
backlog |
high |
nw-whey-cc |
2026-05-29 02:28 |
| #567 |
llmmsg-srv |
└─ Land canonical memory-discipline.md in sh.git (docs/md), commit+push; nw-* pull propagates to 3 hosts |
backlog |
high |
bin-whey-cc |
2026-05-29 01:56 |
| #566 |
llmmsg-srv |
└─ Consolidate evolutiva-memory-discipline.md into fleet-neutral canonical + scrub project/agent names for public sh.git |
backlog |
high |
nw-whey-cc |
2026-05-29 01:56 |
| #570 |
llmmsg-srv |
└─ Acceptance gate: fresh agent on each host must cite the policy+thresholds when asked is-there-a-md-policy |
backlog |
normal |
pm-llmmsgsrv-cc |
2026-05-29 02:28 |
| #571 |
llmmsg-srv |
└─ memory-discipline.md threshold divergence: prose (§10/§13/§15 ~4k tok/150-200 lines/8000 budget) != enforced memlint-cron.sh (600 warn/1200 trim); make doc reference script as threshold SSOT, drop duplicated numbers |
backlog |
normal |
nw-whey-cc |
2026-05-29 02:25 |
| #568 |
llmmsg-srv |
└─ Stub evolutiva-memory-discipline.md -> sh.git canonical; update maintainer-agents.md pointer |
backlog |
normal |
nw-whey-cc |
2026-05-29 01:56 |
| #560 |
llmmsg-srv |
ARO topology v3 fleet-wide rollout (epic) |
backlog |
high |
pm-llmmsgsrv-cc |
2026-05-27 00:33 |
| #564 |
llmmsg-srv |
└─ Phase 6: llmmsg-protocol.md §ARO Membership + §Routing rewrite |
backlog |
normal |
pm-llmmsgsrv-cc |
2026-05-27 00:33 |
| #556 |
llmmsg-srv |
chat-duo-web tablet-grade evolution (path A for unified browser client) |
backlog |
high |
coder-chatduo-cc |
2026-05-26 06:31 |
| #542 |
llmmsg-srv |
Agent-hygiene: automated CLAUDE.md/memory bloat detection + LLM compression (fleet-wide) |
backlog |
high |
pm-llmmsgsrv-cc |
2026-05-25 02:03 |
| #549 |
llmmsg-srv |
└─ chat-duo Agent Hygiene pane: surfaces top-bloat agents from /agent_hygiene |
backlog |
low |
coder-chatduo-cc |
2026-05-23 18:52 |
| #531 |
llmmsg-srv |
Transport reliability + observability program (whey/venus/lezama) |
backlog |
high |
pm-llmmsgsrv-cc |
2026-05-23 18:46 |
| #617 |
llmmsg-srv |
└─ Fleet context-cost rollout: send-keys /compact watchdog + aro_leave-on-idle |
backlog |
high |
pm-llmmsgsrv-cc |
2026-06-03 16:03 |
| #621 |
llmmsg-srv |
│ └─ WI592 nag: drop CronCreate /compact from emit_context_budget_nag interactive branch once send-keys watchdog verified fleet-wide |
backlog |
normal |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #620 |
llmmsg-srv |
│ └─ cc-context-monitor safety-net: flag/extend nag for agents that miss self-leave (seq AFTER rule lands) |
backlog |
normal |
cc-context-monitor-lezama |
2026-06-03 15:28 |
| #625 |
llmmsg-srv |
│ └─ path-B UserPromptSubmit injection copy: reframe CronCreate as automatic-fallback not manual action (avoid conflict w/ v2.13 over-cap DM) |
backlog |
low |
pm-llmmsgsrv-cc |
2026-06-03 16:14 |
| #624 |
llmmsg-srv |
│ └─ cc-context-monitor measured role-cap version (floor+EOQ headroom) - successor to #617 seed caps |
backlog |
low |
bin-whey-cc |
2026-06-03 16:14 |
| #614 |
llmmsg-srv |
└─ Idle over-budget agents never self-compact: wire host-side actuator + fix 3 monitor bugs |
backlog |
high |
bin-whey-cc |
2026-06-02 03:25 |
| #615 |
llmmsg-srv |
└─ chat-duo-web 9704 to lezama: move off whey-initiated -R (VPN-dependent) to lezama-initiated -L over public route |
backlog |
normal |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #539 |
llmmsg-srv |
└─ Venus-side watchdog + alert standardization (mirror lezama pattern) |
inProgress |
normal |
nw-venus-cc |
2026-05-29 05:30 |
| #540 |
llmmsg-srv |
└─ Shim transport prefer-list: loopback -> ZT -> public DNS (defer until W3+W4 prove design) |
backlog |
low |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #320 |
llmmsg-srv |
multi-subscriber SSE + sender echo + client dedupe (chat-duo + chat-duo-web cohabit) |
backlog |
high |
llmmsg-srv-cc |
2026-05-11 22:07 |
| MSG-51 |
llmmsg-srv |
Triage edge LLMMSG_HUB_BEARER rotation: exposed on GitHub via api-images repo (NOT llmmsg-only exposure, so no-rotate rule does not apply) |
backlog |
normal |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| MSG-49 |
llmmsg-srv |
Unknown /command errors back to sender (not silent-send); usage-error on malformed known verb |
backlog |
normal |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| MSG-23 |
llmmsg-srv |
hub.mjs latent SQLITE_ERROR: 'no such column: ""' at Database.prepare (seen 00:47:15 pre-restart). Route interpolates empty string into query as = "" (SQLite parses "" as column id). Find the unparameterized query, convert to bound params. Route-triggered, not fixed by restart. Also tidy hub.mjs:1007 error string to match widened regex. |
backlog |
normal |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #634 |
llmmsg-srv |
Durable GC for stale bridge registrations (TTL prune) - no more retired-codex corpses in registrations.json |
deferred |
normal |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #616 |
llmmsg-srv |
Push llmmsg protocol mechanics into MCP send tool + hub (shrink model-held context) |
backlog |
normal |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| MSG-46 |
llmmsg-srv |
cdw PWA: make chat-duo-web installable (manifest+icons+service worker, Add-to-Home-Screen) |
backlog |
normal |
coder-chatduo-cc |
2026-06-15 02:51 |
| MSG-9 |
llmmsg-srv |
venus /opt/llmmsg-srv is NOT a git checkout - no sync path for maintainer pushes (forces manual /gdrive/temp hops; v1.16 bootstrap had to be hand-installed). Make it a proper git checkout of origin/main like lezama (project_lezama_code_drift) so future pushes land via git pull. Care: venus dir has live local files (.agent-name, service config) - clone-in-place or backup+clone, don't clobber runtime config. |
backlog |
normal |
— |
2026-06-06 04:56 |
| #623 |
llmmsg-srv |
gitmerge.sh diverged-branch newest-wins auto-adds untracked files, defeats gitpush --only strict-staging |
backlog |
normal |
bin-whey-cc |
2026-06-03 16:14 |
| #606 |
llmmsg-srv |
lezama single-path fragility: all 5435 probes since May23 are public-dns with zt_whey_visible=0 (ZeroTier fallback never worked); make ZT a real 2nd route + fix phantom ssh_r_active self-report ↳#608 |
backlog |
normal |
nw-whey-cc |
2026-06-02 05:00 |
| #605 |
llmmsg-srv |
lezama host-dark alert: watchdog fires notify-elazar+eq after 3 consecutive 'unreachable' skips (~15min), instead of silently skipping forever ↳#608 |
backlog |
normal |
nw-whey-cc |
2026-06-02 00:45 |
| #593 |
llmmsg-srv |
/loop skill self-exports LLMMSG_AUTONOMOUS=1 on entry (so L1 hook branches correctly inside /loop) |
backlog |
normal |
pm-llmmsgsrv-cc |
2026-05-31 13:29 |
| #463 |
llmmsg-srv |
Next.js chat client to supersede chat-duo GTK + chat-duo-web (brainstorm item 4): node service on whey, Tailscale-only, native Pending tabs. Needs design brainstorm->spec->plan. |
backlog |
normal |
coder-chatduo-cc |
2026-05-26 08:37 |
| #529 |
llmmsg-srv |
chat-duo timestamp display drops date - stale msgs read as today |
backlog |
normal |
coder-chatduo-cc |
2026-05-23 18:07 |
| #524 |
llmmsg-srv |
aro_config caller field unauthenticated - light-scope hardening |
backlog |
normal |
— |
2026-05-23 06:20 |
| #438 |
llmmsg-srv |
cc-context-monitor compact half - wire host-side tmux send-keys auto-compact for idle over-cap sessions, after the one-session verification test |
backlog |
normal |
— |
2026-05-20 15:47 |
| #398 |
llmmsg-srv |
hub-resident cross-project WI board: add cross_wi table + MCP wi_add/wi_close/wi_ls/wi_show/wi_event tools. Replaces per-host bwi for cross-project items (mars/pluto/venus/evolutiva). 3-PM consensus pm-mars-29395 / pluto-29405 / venus-29411; Elazar greenlight elazar-…-29506 Q2. Owner: llmmsg-srv-cc-w. |
backlog |
normal |
— |
2026-05-17 05:04 |
| #158 |
llmmsg-srv |
Brainstorm: connect lezama agents to whey's llmmsg-srv hub |
backlog |
normal |
— |
2026-05-02 03:38 |
| MSG-73 |
llmmsg-srv |
cdw: Caddy flush_interval -1 for SSE reconnecting issue (Caddyfile) |
backlog |
low |
nw-venus-cc |
2026-06-19 17:54 |
| MSG-50 |
llmmsg-srv |
dm-* send-authorization: reject non-owner/non-host-nw/non-Elazar sends to or origin_aro'd-as aro:dm-* |
backlog |
low |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #597 |
llmmsg-srv |
Hub cleanup traps (post-#594): systemd unit sets LLMMSG_SRV_DB but code reads LLMMSG_DB (no-op env var); stale llmmsg.sqlite + init-db.sh schema conflicts with live v2.sqlite |
backlog |
low |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| #596 |
llmmsg-srv |
Refresh venus /opt/llmmsg-srv flat copy from whey (hub.mjs+shim) - stale May-18 scp caused misdiagnosis; do NOT delete (venus MCP shim runs from it) |
backlog |
low |
coder-llmmsgsrv-cc |
2026-06-15 08:39 |
| MSG-47 |
llmmsg-srv |
MSG-30 lezama: coordinated full-tree pull-forward (shim v2.9.47 + bootstrap v1.19 + package.json, atomic) - replaces nw-lezama hand-patch, un-dirties staged v1.18 bootstrap |
backlog |
low |
pm-llmmsgsrv-cc |
2026-06-15 03:37 |
| MSG-43 |
llmmsg-srv |
Rationalize errscan-vs-applog overlap fleet-wide (both = error->PM-DM; applog covers all 3 proj+now green, errscan mars-only, mars DOUBLE-covered) - decide consolidate-not-expand; needs PM/Elazar division-of-labor call |
deferred |
low |
pm-llmmsgsrv-cc |
2026-06-14 15:04 |
| MSG-26 |
llmmsg-srv |
llmmsg-kpi-report.sh: strip dead -small ARO branch from §PM .aro action list (lanes retired MSG-17) |
backlog |
low |
pm-llmmsgsrv-cc |
2026-06-10 04:04 |
| #647 |
llmmsg-srv |
.llmmsg-aros cleanup: remove 16 repo-tracked copies in pensanta website repos (#631 tail) ↳#631 |
backlog |
low |
nw-venus-cc |
2026-06-05 04:38 |
| #607 |
llmmsg-srv |
lezama out-of-band reset: a dead kernel cannot self-recover; scope IPMI/iDRAC vs smart-PDU for remote host-dark power-cycle (blocked on Elazar: does box have IPMI?) ↳#608 |
backlog |
low |
pm-llmmsgsrv-cc |
2026-06-02 00:45 |
| #465 |
llmmsg-srv |
Document to=pm:X send target + hub PM election/pending-queue in README.md protocol section (shipped in 5741b17, README not yet updated) |
backlog |
low |
llmmsg-srv-cc |
2026-05-22 03:53 |