- acme-dns.nix: ONE dual-zone SAN cert (ci+*.ci.autonomic.zone AND ci+*.ci.commoninternet.net)
via the same acmedns account — storage re-keyed by cc-ci-acme-storage-seed.service; handoff
reads the new cert dir. Single secret pair => zero changes to the traefik reconciler.
- dashboard/bridge/reports: dual Host rules during the bake window (bridge gets explicit
parentheses so && does not shadow the dashboard on the new host).
- drone abra app renamed to drone.ci.autonomic.zone (fresh DB supported: DRONE_USER_CREATE
re-injects the sops bridge token); runner RPC + bootstrap-drone-oauth.sh follow.
- harness: app_domain() issues *.ci.autonomic.zone run domains; RUN_APP_RE / stack-name
regexes / docker-prune accept BOTH zones during the bake. Warm stacks deliberately stay
on the legacy zone (data-warm volumes; post-bake migration).
- URLs in bridge/dashboard defaults + recipe-report.py follow the new names.
The operator enabled China-hosted models on the workspace, so deepseek-v4-flash
is available on the Go tier again (it was the model the weekly run used before
the host moved off ZEN, and it is the cheap/fast one for the per-recipe
subagents). Verified on the host: `opencode run --model
opencode-go/deepseek-v4-flash` answers.
The weekly run's MAIN agent moves to opencode-go/glm-5.3-flash separately, in
upgrader.env on the host.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FqkQq3CDmFWcQ7u1LzoyRz
The host's opencode credential is an OpenCode Go subscription key, so ZEN
models (`opencode/…`) fail there with "Insufficient balance". The weekly
run's recipe subagents read this file, so they move to the Go provider.
glm-5.2 rather than deepseek-v4-flash because the latter is China-hosted on
Go and needs a per-workspace opt-in.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FqkQq3CDmFWcQ7u1LzoyRz
cc-ci-secrets now encrypts to the new host (195.201.88.249) via its own
ssh-host-key-derived age identity, like the canonical cc-ci did, so the
off-box master recovery key no longer has to live on that box —
/var/lib/sops-nix/key.txt there holds the host-derived identity instead.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FqkQq3CDmFWcQ7u1LzoyRz
statix flagged the repeated `systemd.` keys (tmpfiles marker, acme-dns
daemon, traefik handoff oneshot); they are now one nested attrset. Purely
structural: `#cc-ci` still evaluates. With #33 this makes the push
self-test's lint stage pass again.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FqkQq3CDmFWcQ7u1LzoyRz
`scripts/lint.sh --fix` from the pinned lint devshell: 90 Python files
reformatted (ruff format, mechanical) and one C420 (dict comprehension →
dict.fromkeys) in tests/unit/test_f211_sso_skip.py. The push self-test had
been failing at the lint stage since build 1313 (2026-08-31) on exactly
these files; nothing else changed.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FqkQq3CDmFWcQ7u1LzoyRz
The whole server (every service module, the harness tooling, sops wiring,
acme-dns) becomes one reusable module, nix/modules/default.nix, so another
flake can run cc-ci on a host it defines. First consumer: the
cc-ci-orchestrator repo's `#cc-ci` host, which runs the CI server and the
orchestrator together on one Hetzner machine.
Two things the modules hard-coded become options (nix/modules/options.nix):
- cc-ci.publicIPv4 — acme-dns's listen address and ns-acme glue record.
- cc-ci.sopsFile — the secrets.yaml path; defaults to the secrets/ submodule,
but a consumer that imports cc-ci as a plain input (no private submodule)
points it at the deployed --recursive checkout and sops-nix reads it at
activation (validateSopsFiles off for that case).
The standalone host (nix/hosts/cc-ci-hetzner) now only carries hardware,
networking and identity and imports the module via the flake. Verified: the
`#cc-ci` system derivation is byte-identical before and after
(/nix/store/ckp1244bz86fz3qbx81n5kx60c1lak3m-…531670d.drv on both).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FqkQq3CDmFWcQ7u1LzoyRz
Route deepseek-v4-flash through the opencode zen subscription endpoint
(opencode/deepseek-v4-flash) instead of tinfoil pay-per-use. The zen
subscription is already paid for; flash consumes the balance far more
slowly than glm-5.2 did (flash cost ~$0.0000002/tok vs glm's higher rate),
so a full weekly run is much less likely to exhaust the balance — and
the operator has reset it.
Completes the fleet-wide move: upgrader parent (LOOP_MODEL) + report
(REPORT_MODEL) in upgrader.env, and this PR sets the subagents. Supervisor
stays on opencode-go/glm-5.2.
Verified: zen endpoint serves deepseek-v4-flash (HTTP 200, 1.2s).
Switch the general subagent model from opencode/deepseek-v4-pro (zen
endpoint, subject to workspace balance limits) to tinfoil/deepseek-v4-flash
(pay-per-use API, no rolling balance limit, cheaper than pro).
This completes the fleet-wide move to deepseek-flash on tinfoil: the upgrader
parent (LOOP_MODEL) and report (REPORT_MODEL) are set in upgrader.env, and
this changes the subagents. The hourly supervisor stays on glm-5.2.
Verified: tinfoil endpoint accepts deepseek-v4-flash (HTTP 200, 2.3s response).
Trial run next Friday (2026-08-21) to evaluate flash capability on the
upgrade workload; fall back to pro if it struggles on complex recipes.
autonomic-bot
notplantsandClaude Opus 5
autonomic-bot