CHANGELOG
Power Agents Evolved — What's new.
More Savings. Faster Workflows. New Innovations — Shipped weekly, tagged by impact — throughput, stability, analytics, UX.
Shipped continuously… Throughput wins: more seats, smarter routing.
- Throughput wins: more seats, smarter routing.
- Stability: recover, rewake, heal stalled work.
- Analytics: know burn before you hit the wall.
- UX: dashboards, badges, timelines, focus fixes.
- Tagged by impact — skim what matters to you.
- Weekly cadence: real ships, clear notes.
Publish typecheck
Fixed
- Publish typecheck
Kimi hero copy; FlowCounts codeReviews/requests/attempts; simpleFlow validation cast.
Release notes
Maintenance release.
Release notes
Maintenance release.
Release notes
Maintenance release.
Release notes
Maintenance release.
Release notes
Maintenance release.
HostUsagePort grant/window capabilities
Added
- HostUsagePort grant/window capabilities
hasQuotaWindows, supportsUsageGrants, supportsAutoApplyExpiringGrants + hostUsageCapabilities() so Apply / auto-apply chrome is interface-driven (Claude true; Grok/Codex honest false until APIs exist). Docs: docs/features/host-usage-grant-capabilities.md.
Extra usage on /api/oauth/usage
Added
- Extra usage on /api/oauth/usage
expose real extra_usage (no longer always null). Unused extra-usage / reset credits that expire are auto-applied 5 minutes before expiry (default ON; powerClaude.proxy.autoApplyExpiringExtraUsage). Docs: docs/features/extra-usage-auto-apply.md.
- Host account Pin
(Grok / Codex / Gemini / …) — Profiles + pc host-account pin <host> <id> stay attached to a seat so leftover usage is drained before it expires. Rotate/assign honor pins and fall back if the pin is limited.
- Reset burn-down is more aggressive
quadratic urgency now ramps over the full 5h / 7d window (not only the last 90 minutes) and uses leftover-at-reset from past burn rate so usage is not left on the table.
- Power Claude Council
(ability_id=agent-debate): N-agent decision engine with independent analysis, Claim Ledger, evidence, dissent-preserving Decision Packets, Inspect Mode, CLI (pc council / pc debate / pc red-team), and VS Code commands. Two-agent debate is a preset. Dev-gated. See docs/architecture/council.md.
- Multi-host marketplace positioning
package description, capabilities SSOT, and keywords lead with Model tabs for Claude / Grok / Codex / Gemini / Kimi (by popular demand), without changing money dual-trial claims.
Fixed
- Command Center 3.0 CAPACITY
dual usage meters enrich from rate-limits SSoT when fleet util is null (no more all "no probe yet" while probes exist); denser reset chrome + per-window hints + How resets work list; Suggestions strip (recommend.json) deep-links to Usage · recommendations; Events L1 opens structured Home Events tab (raw jsonl remains palette "Open Events Log").
- Dedupe
still inject-only accountsHtml (shared usageWindowsHtml + scopes toolbar + suggestions strip); client does not remake usage rows.
Stop losing afternoons to rate limits.
Premium Trial: card-on-file, $0 today — locks today's founder rate at $12/mo for 2 years. Free: no card, 7-day trial inside the extension.
Neural-LLM is an independent software company. We are not affiliated with, endorsed by, or sponsored by Anthropic, PBC, OpenAI, xAI, or Moonshot AI. Power Agents products are third-party tooling. Claude™, Codex™, Grok™, and Kimi™ are trademarks of their respective owners, used here only to identify the software each product works with.