What problem does it solve? When returning to a project after a gap or starting a session cold, it is hard to answer "how far along am I and when will it be done?" without manually digging through git history, plans, and checklists. This Skill computes that status read fresh on every call, with honest provenance tagging so you know which numbers are attested facts and which are guesses. ## Core Features & Use Cases - Position and remaining-work read: Gathers evidence from git (branch, dirty count, ahead/behind, commits), plan files (open/done checklist counts, current phase), and folder signals via a bounded read-only subagent, returning only a distilled bundle. - Banded ETA with fail-closed honesty: Forecasts are always ranges tagged "estimate, not observed"; on barren evidence the output degrades visibly (gate_status degraded/indeterminate) instead of fabricating a confident answer. - --resume mode: Rebuilds context after a break with exactly one next move, its counter-case, and the dropped alternative. - Three output faces: A structured YON record for agent consumers, a plain-language markdown read for humans, and a visual widget (Claude Code) with a mandatory information-complete ASCII twin. - Use Case: You return to a feature branch after a week away and ask where things stand; the Skill reports the current phase, 4 open items out of 5, a 1-3 day ETA band, and the single next move to pick the thread back up. ## Quick Start Ask the agent "where are we on this repo and what's the ETA?" or invoke /orient-status, optionally passing a one-line goal so distance and ETA fields can be computed.
