Step 1 of the locked auth-foundation-first build order, and the keystone of the whole roadmap.
Turns the Project Board from a read-only viewer into a command surface: one click fires a ROADMAP.md cc-item at the fleet as a real mission, which loops back onto the board.
- Specced as Option A (write-back + daemon-wide loopback auth), 3 critique rounds. Not built.
- Spec:
docs/superpowers/specs/2026-07-06-local-project-tracker-design.md
- Local spec branch:
docs/local-project-tracker-spec
Why this is first
Phase 2 and Remote Control share one security foundation. fleetd today is an unauthenticated daemon on 127.0.0.1:8787 holding the operator's ambient gh/git credentials (full push) — it cannot be exposed remotely as-is. Phase 2 already forces router-level loopback auth plus a token handshake over every mutating endpoint (tracker spec §7.4), which is exactly Remote Control's prerequisite. Building Phase 2 first buys down remote-control risk instead of paying for auth twice.
Blocks the Remote Control spec.
Step 1 of the locked auth-foundation-first build order, and the keystone of the whole roadmap.
Turns the Project Board from a read-only viewer into a command surface: one click fires a
ROADMAP.mdcc-itemat the fleet as a real mission, which loops back onto the board.docs/superpowers/specs/2026-07-06-local-project-tracker-design.mddocs/local-project-tracker-specWhy this is first
Phase 2 and Remote Control share one security foundation.
fleetdtoday is an unauthenticated daemon on127.0.0.1:8787holding the operator's ambientgh/git credentials (full push) — it cannot be exposed remotely as-is. Phase 2 already forces router-level loopback auth plus a token handshake over every mutating endpoint (tracker spec §7.4), which is exactly Remote Control's prerequisite. Building Phase 2 first buys down remote-control risk instead of paying for auth twice.Blocks the Remote Control spec.