The launchagent and ansible run without mise activation, so a bare cargo/rustc
shim falls back to rustup's default toolchain — which lagged heph's rust-version
floor (1.89) on both indri (1.87) and gilbert (1.84), silently failing the build.
Pin the channel explicitly in the bootstrap env and the plist.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Deploy hephd --mode server on indri as a self-updating LaunchAgent managed
by Ansible (ansible/roles/heph, tag heph), making indri the canonical heph
hub for the hub-and-spoke task/context system.
- Server mode on 0.0.0.0:8787, self-update every 10 minutes (cargo install
from the public forge URL; ~/.cargo/bin on the agent PATH).
- heph-pwa shell served via --web-root straight from a version-pinned checkout,
TLS-terminated at heph.ops.eblu.me through Caddy (new caddy_services entry).
- New Authentik device-code (RFC 8628) OIDC app 'heph' (public client) plus a
default-device-code-flow bound to the default brand's flow_device_code.
- Docs: new services/hephaestus.md service card (incl. Path A seeding runbook
and the gilbert spoke caveat), indri.md service list, changelog fragment.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
