Changelog

Old /pricing rendered three plan cards from JS-fetched site settings — visitors saw a "Loading plans…" flash before anything rendered. Replaced with a server-rendered four-tier card grid + a feature comparison matrix + an interactive ROI calculator + 8 FAQ entries.

• Free $0 — 25 runs / month on cloud models we pay for. No Ollama install required.
• Pro $39 / month — 500 runs / month, full model lineup, multi-file refactor mode, test-first mode, BYO API keys.
• Team $59 / seat / month (min 5 seats) — org workspace, shared skills, role-based access, 99.5% SLA, 365-day audit.
• Enterprise from $25k / year — on-prem option, SSO, SCIM, EU residency, dedicated CSM.

Every plan includes the no-training guarantee, permission-scoped sandbox, and replay. Pricing copy is now indexable; sitemap updated to include /pricing and /download.

Trust appears in product, not just on landing. Shipped:

• New public /security page with the four trust pillars (no-training guarantee, permission-scoped sandbox, verified output, reversible by default).
• Privacy and Terms updated with the explicit no-training contractual commitment.
• "No training" badge in the chat header, visible on every product surface.
• Public security.txt at /.well-known/security.txt per IETF RFC 9116 — vulnerability reports go to [email protected].
• Header / footer trust nav: Security & Trust → Privacy → Terms.
• Sitemap updated.

Internally we ran a brutal positioning audit and locked the public direction for the next 90 days. Highlights:

• New tagline: It writes the code. It runs the build. It fixes what broke. Replaces the category-empty "AI that ships software."
• Pricing restructured to four tiers (Free / Pro / Team / Enterprise) — Pro anchored at $39/month to signal real engineering work, not autocomplete.
• Internal "10 phases" engineering vocabulary retired from public surfaces. Customer-facing language now leads with verified output, .NET-native, and permission-scoped trust.
• Roadmap restructured around outcome themes — Trust → Distribution → Depth → Reach → Intelligence — instead of phase numbers.
• Forward path through the next 90 days locked in docs/eraevo/EXECUTION_TRACKER.md: 100+ paid Pro users, $10k+ MRR, VS Code extension and GitHub App live, first design partner.

Public docs landed at docs/eraevo/ in the source repo (≈ 4,800 lines).

Before this changelog began, the EraEvo orchestrator was built end-to-end over 2026-Q1: execution sandbox, AI provider routing (4 providers), self- healing build loop (15+ C# error codes), task planning, multi-file generation with consistency checking, hybrid local/cloud routing, test generation + execution, execution memory + strategy optimization, Roslyn AST + semantic index, multi-agent debate (Architect / Developer / QA / Reviewer). Backed by 224 unit tests, all green.

Web app at eraevo.com, Core API behind Caddy at core.eraevo.com, background Worker, operator Dashboard, and Windows desktop client (WPF Prism + DryIoc + WebView2) shipped with MSI installer and FlaUI E2E tests.