Continues: #127 (ecosystem adoption v1)
Goal
Make Gittensory the gittensor-native quality + planning layer that a contributor's own AI harness (Claude Code, Cursor, Codex) cannot replicate, and surface it where people already work. The wedge is a shared, deterministic "real contribution vs AI slop" definition: help miners ship real PRs and help maintainers enforce quality — the #1 maintainer complaint about Gittensor work. We deliberately do not try to be the code agent.
Pillars
Related existing work (reuse, do not duplicate)
Done when
- A miner can self-check slop risk locally and receive the gittensor rubric before opening a PR.
- A maintainer can enable a slop gate + per-line check annotations and view quality trends.
- The browser extension serves both miners and maintainers, and a miner is alerted the moment a review requests changes on their PR.
Cross-cutting acceptance criteria
npm run test:ci passes; 97%+ global coverage. No source upload from miner tooling (fail closed).
- Public output never exposes wallets, hotkeys, reward estimates, raw trust scores, public score predictions, private reviewability/scoreability, or farming language.
- Advisory-only / human-in-the-loop preserved: gates default to advisory, never auto-merge or auto-submit.
UI evidence gate
Any visible web, browser-extension, or GitHub-overlay change must include maintainer-reviewable screenshots or a short recording covering the changed states. A checked template box without actual visual evidence is not enough.
Wave 1 — remaining after the 2026-06-17 discovery sweep → Milestone M1 (due 06-22)
Backend is largely shipped; remaining work is dashboard surfaces + write-safety bugs + email. Detail: docs/roadmap/discovery-sweep-findings.md.
Next wave
⭐ Immediate focus → #805 — PMF: non-registered → gittensor onboarding funnel (Milestone M0, due 06-20)
The prioritized front of Wave 1: prove the onboarding funnel on awesome-claude / gittensory / metagraphed before scaling. Children #799–#804.
🔄 Upstream sync → #814 — gittensor scoring-sync hardening + parity
Keep scoring/outcomes/planning in parity with entrius/gittensor. 2 correctness bugs to M0 (#806/#807); parity+hardening #808–#812; upstream watch #813 (incl. gitt miner advisor #1460).
Continues: #127 (ecosystem adoption v1)
Goal
Make Gittensory the gittensor-native quality + planning layer that a contributor's own AI harness (Claude Code, Cursor, Codex) cannot replicate, and surface it where people already work. The wedge is a shared, deterministic "real contribution vs AI slop" definition: help miners ship real PRs and help maintainers enforce quality — the #1 maintainer complaint about Gittensor work. We deliberately do not try to be the code agent.
Pillars
Related existing work (reuse, do not duplicate)
MAX_LIST_ITEMSsafety cap #522 / [Bug]: Pre-auth rate limits collapse tounknown-ipwhencf-connecting-ipis absent, so one client can throttle everyone behind fallback proxies #477 / feat(issues): respect declined (closed) contributor-issue drafts to stop re-proposal churn #517 stability + correctness hardeningDone when
Cross-cutting acceptance criteria
npm run test:cipasses; 97%+ global coverage. No source upload from miner tooling (fail closed).UI evidence gate
Any visible web, browser-extension, or GitHub-overlay change must include maintainer-reviewable screenshots or a short recording covering the changed states. A checked template box without actual visual evidence is not enough.
Wave 1 — remaining after the 2026-06-17 discovery sweep → Milestone M1 (due 06-22)
Next wave
⭐ Immediate focus → #805 — PMF: non-registered → gittensor onboarding funnel (Milestone M0, due 06-20)
The prioritized front of Wave 1: prove the onboarding funnel on awesome-claude / gittensory / metagraphed before scaling. Children #799–#804.
🔄 Upstream sync → #814 — gittensor scoring-sync hardening + parity
Keep scoring/outcomes/planning in parity with entrius/gittensor. 2 correctness bugs to M0 (#806/#807); parity+hardening #808–#812; upstream watch #813 (incl. gitt miner advisor #1460).