4f177f9b8d
* wip(hermesworld): viral sprint checkpoint - landing rebuild + character pipeline scaffold - standalone /hermes-world and /world routes bypass workspace shell - root overlay leaks gated for landing + game surfaces - character pipeline scaffolding (player/npc/glb-body components) - canonical asset path public/assets/hermesworld/characters/ - docs: landing-page-spec, graphics-usability-plan, agora-believable-checklist, master-roadmap - handoff at memory/goals/2026-05-05-hermesworld-viral-sprint/handoff.md Local-only checkpoint. Not for upstream yet. * feat(playground): persistent admin mode toggle with shield button - Admin mode now persists via localStorage (key: hermes-playground-admin) - Shield icon button in HUD (right rail, below focus toggle, md+) - Click toggles admin panel and saves preference - ?admin=1 URL param still works as override - gitignore swarm worker scratch dirs Mission: memory/swarm/missions/2026-05-05-pr-triage.md (5 swarm lanes dispatched on 19 open PRs, no-merge contract) * feat(landing): add Play Now CTAs to HermesWorld landing - Hero: Play Now (primary, gold), View on GitHub (demoted), Read Roadmap - Header nav: Play badge (highlighted gold) - Final CTA: Play Now (primary), GitHub + Roadmap (secondary) All Play buttons go to /playground which mounts the title screen (username + character customizer + Enter). Sets up the public-URL deploy: hermes-world.ai → / serves landing → click Play → /playground. * fix(tasks): use shared kanban backend --------- Co-authored-by: Aurora release bot <release@outsourc-e.com>
1.7 KiB
1.7 KiB
PR Triage Mission — 2026-05-05
Repo: outsourc-e/hermes-workspace Goal: triage all 19 open PRs, produce structured findings per PR, NO AUTO-MERGE. Reporter: Aurora (main session) Lead time: within this session (~30-45 min for first reports back)
Hard rules
- Do NOT merge any PR. Eric must approve every merge personally.
- Do NOT push to main or any contributor branch.
- Do NOT close PRs. Just classify them.
- Report back to Aurora with a single concrete checkpoint per PR.
- If you encounter conflicts, list them; do not resolve them on the PR branch.
Per-PR deliverable
For each assigned PR, produce one structured block:
PR #<num> — <title>
AUTHOR: <login>
CLASSIFICATION: APPROVE_READY | NEEDS_FIX | NEEDS_DISCUSSION | CLOSE_RECOMMEND
DIFF_SUMMARY: <2-3 lines, what it actually changes>
FILES_TOUCHED: <count + key paths>
RISK: low | medium | high — <one-line reason>
TESTS_RUN: <pnpm build / pnpm lint / specific test files / N/A>
TEST_RESULT: pass | fail | not-run — <evidence>
CONFLICTS: none | <files>
BLOCKERS: <or none>
REVIEW_NOTES: <2-4 lines — concrete things Eric should know before merging>
RECOMMENDED_ACTION: merge | request-changes:<what> | close:<why> | discuss:<what>
Lane assignments
Healthy workers and their PR slices (by PR number):
- swarm3 → 338, 336, 335, 334
- swarm4 → 332, 330, 327, 325
- swarm6 → 324, 322, 320, 318
- swarm10 → 317, 315, 310, 308
- swarm12 → 307, 301, 299
Quarantined: swarm1 (broken runtime), swarm2/swarm11 (rate-limited 429).
Checkpoint cadence
Workers must produce one checkpoint per PR and report it back via STATE: HANDOFF blocks. Aurora will collect, summarize for Eric, and Eric decides each merge.