Blueprint v2 — Re-verification (Tam, research lane)

Re-dispatch of the final audit (Atlas → Tam). Re-probed live 09:00 AEST 2026-09-22. Read-only; nothing modified on server. Prior audit = 2026-09-22-blueprint-v2-audit (08:55 snapshot). This note confirms that verdict and records one fresh delta.

1. Tunnel — RECONNECTED ✅

• ssh pid 843761 (Sep 21 20:15) ESTABLISHED to 100.124.146.127:22.
• Forwards live: 18300-18306 → 200; costgov 8430 ok; registry 8432 ok; Tam A2A 8447 alive.
• DELTA (positive): Tam now registered in 8432 (id=tam, port=8447, last_seen 33s, fresh).

2. Blueprint doc — UNCHANGED, discrepancies persist

• Doc = /home/ubuntu/.solos/docs/blueprint-unified-solos-implementation.md (18,790 B, Sep 21 21:14). Atlas's laptop path (~/.solos/docs/) is STILL wrong — file is server-side only.
• Future-dated: header "Date 2026-09-24" (today 09-22).
• Self-contradictory: "LOCKED by Young (09-18)" vs §8 approval PENDING.
• Weighting drift: header "OpenCode 60 / Sol-custom 30 / Hermes 0%" (sums 90) vs Young-locked 60/30/10.

3. Live agent health (probed 09:00)

• 8432 (central) 11 agents: ace/alpha/atlas/echo/eva/mira/nova/sol/tam online; eve down; min flagged down.
• ⚠️ B4 CONFIRMED (fresh instance): min (8307) registry="down" is FALSE — run_min.py alive 26+ min, /health → 200 in ~50ms via tunnel (3/3). Registry last_seen frozen at 1790031329 (age 268s and growing) → heartbeat not refreshing despite healthy serve.
• eve (8442) genuinely DOWN: eve-a2a.service + eve-a2a-bridge.service auto-restart crash-loop, ExecStart still /home/ubuntu/.hermes/.../server.py (status 200/CHDIR, removed Hermes path). B2 unfixed.

4. Blockers — current status

• B1 TG gateway: still absent (no gateway_loop.py, no telegram/gateway unit). CRITICAL.
• B2 Eve crash-loop on removed ~/.hermes/ path. Unchanged.
• B3 Double-registration: 8432 vs 8433 STILL divergent (8433 has "eva-laptop", 8432 has "eva"). Note: laptop local-8433 forward now aliases → server-8432, masking divergence from the laptop.
• B4 Registry status untrustworthy: batch last_seen (all fresh=True incl. down agents); min healthy-but-"down" is the concrete case.
• B5 Overlapping tunnel managers: autossh 299169/299170 (Sep12) + ssh 843761 (Sep21) still coexist; stale autossh pair still looping.
• B6 Approvals pending: doc LOCKED-vs-PENDING + future-dated blocks Young sign-off.
• B7 Vault path: laptop-side FIXED + verified (30m cron OK pushed=1 @08:56, RBASE=vault/solos/research/tam). Server-side solos-tam-vault-perms.service Description still says "Research/tam/ drops" (stale).

5. Verdict

NOT ready for sign-off — unchanged. Re-verification at 09:00 confirms every blocker from the 08:55 audit still stands; nothing regressed and nothing newly fixed server-side. The single critical reachability blocker (B1 Telegram gateway) remains absent, and the registry's status field is freshly proven wrong (min healthy but marked down).

🎯 Tam's Read: Tunnel is up and the vault sync is healthy, but the verdict holds — Telegram gateway still missing, Eve still crash-looping on the deleted Hermes path, and the registry now mislabels a demonstrably healthy min as down; nothing server-side has moved since 08:55.