Compare Mindaeon against Claude Code, Cursor 2.0, Devin 2.2, Factory, GitHub Copilot, and GitLab Duo across air-gapped sovereignty, AST mutation gates, hybrid code intelligence (71.2% fewer tokens / 3.5x vs grep), and FinOps.
Evaluated across air-gapped sovereignty, deterministic pre-merge verification, multi-repository contract safety, hybrid code intelligence, FinOps cost control, and compliance provenance. All columns use identical capability indicators (✓ Built-in, ◐ Partial, ✕ Not included).
| Capability / Architectural Guarantee | Mindaeon (v0.1.0) | Claude Code | Cursor 2.0 | Devin 2.2 | Factory | GitHub Copilot | GitLab Duo |
|---|---|---|---|---|---|---|---|
| Sovereignty True Air-Gapped Single-Host Appliance | ✓ Built-in 1,091 MiB idle RAM across all 8 services (Keycloak, PostgreSQL 18, Gitea, SeaweedFS, Control Plane, OpenBao, Edge, OPA) | ✕ Not included Cloud-only control plane; requires outbound Anthropic/cloud API connectivity | ✕ Not included Proprietary SaaS backend required for indexing, routing, and Background Agents | ✕ Not included Cloud-hosted Firecracker VMs orchestrated by Cognition SaaS | ✕ Not included Multi-tenant SaaS orchestration plane | ✕ Not included GitHub Enterprise Cloud required for Copilot Coding Agent | ◐ Partial Self-hosted model gateway available, requires multi-node GitLab Ultimate |
| Sovereignty Zero Outbound Telemetry by Default | ✓ Built-in 0 external calls; 100% local hash-chained audit logs & OpenBao transit encryption | ✕ Not included Telemetry & usage metadata transmitted to vendor cloud | ✕ Not included Privacy mode still routes requests through Cursor cloud proxy | ✕ Not included Session transcripts and trajectories stored in Cognition cloud | ✕ Not included Cloud session state and analytics retained by vendor | ✕ Not included Requires continuous GitHub Cloud license & telemetry check-ins | ◐ Partial Configurable on self-managed instances with air-gapped license |
| Verification Host-Enforced Test Integrity Guard | ✓ Built-in 100% (7/7) interception of test deletion, .skip/.only, and assertion weakening outside sandbox | ✕ Not included Agent shares write access to test files inside working tree | ✕ Not included Agent can modify or skip failing tests unless manually reviewed | ✕ Not included Full VM write access allows modifying test suites to pass CI | ◐ Partial Prompt-level guardrails without host-enforced AST diff verification | ✕ Not included Relies on downstream PR review to catch weakened tests | ✕ Not included No AST-level test tampering detector |
| Verification Changed-Surface AST Mutation Testing | ✓ Built-in Deterministic AST mutants verified in p95 = 3.77 ms (100% weak-test rejection) | ✕ Not included Not included | ✕ Not included Not included | ✕ Not included Not included | ✕ Not included Not included | ✕ Not included Not included | ✕ Not included Not included |
| Multi-Repo Cross-Repo Contract Radar & Atomic Rollback Sagas | ✓ Built-in 100% breaking-change recall across OpenAPI/GraphQL/Protobuf/TS with atomic rollback sagas | ✕ Not included Single-repository working directory scope | ◐ Partial Multi-root workspace indexing without topological PR saga rollback | ◐ Partial Multi-repo sessions without deterministic contract AST verification | ◐ Partial Cross-repo Droids without cryptographic saga rollback guarantees | ✕ Not included Scoped to single repository per issue/PR workflow | ✕ Not included Scoped to single project merge request context |
| Orchestration Parallel Worktree Collision Radar & Shadow Merge | ✓ Built-in Continuous background dry-run merges cut parallel agent conflicts by 76% (25 -> 6) | ◐ Partial Supports git worktrees manually; no live cross-worktree shadow merge radar | ◐ Partial Up to 8 parallel worktrees; conflicts resolved only at merge time | ✕ Not included Isolated cloud VMs; conflicts surface only after PR creation | ✕ Not included Cloud task branches without real-time AST symbol overlap detection | ✕ Not included Isolated GitHub Actions branches without cross-branch radar | ✕ Not included No worktree orchestration |
| Intelligence Hybrid Code Intelligence (Tree-sitter + SCIP + pgvector RRF) | ✓ Built-in 71.2% prompt token reduction (3.5x fewer tokens than raw grep) and 94% symbol recall via Reciprocal Rank Fusion | ◐ Partial Agentic ripgrep/glob search (high token burn on large monorepos) | ✓ Built-in Cloud-hosted Merkle-tree embeddings + LSP context | ◐ Partial DeepWiki cloud indexing | ◐ Partial Cloud repository graph indexing | ◐ Partial GitHub Blackbird code search + embeddings | ◐ Partial Zoekt exact code search + basic RAG |
| FinOps Deterministic Shadow FinOps & Prompt Cache Optimizer | ✓ Built-in 50.0% verified cost reduction via shadow replay & prefix cache stabilization | ◐ Partial Anthropic prompt caching supported; no multi-provider shadow replay | ✕ Not included Opaque credit-based model routing pool | ✕ Not included Opaque ACU consumption billing | ✕ Not included Standard token/seat billing without shadow model evaluation | ✕ Not included Premium request multiplier quotas | ✕ Not included Fixed add-on seat licensing |
| Compliance SLSA v1.0 / in-toto Provenance & 9 Compliance Packs | ✓ Built-in Ed25519-signed session provenance + SOC2, ISO27001, HIPAA, FedRAMP, PCI, GDPR, DORA, NIS2, EU AI Act | ✕ Not included No built-in SLSA v1.0 per-commit agent provenance | ✕ Not included SOC 2 Type II vendor cert only; no signed diff provenance | ✕ Not included Enterprise audit logs without cryptographic hash chains | ◐ Partial Enterprise audit trails without in-toto attestations | ◐ Partial GitHub audit log events without SLSA agent attestations | ◐ Partial GitLab compliance center integration |
| Surfaces Unified 6-Surface Parity (CLI, Desktop, IDE, Browser, Web, Office 365) | ✓ Built-in 289/289 capability checks passing across Mindcode CLI, Desktop, VS Code, Browser Extension, Console & Office Add-in | ◐ Partial CLI, Desktop, IDE, and Web (no native Office 365 Add-in or self-hosted control plane) | ◐ Partial IDE fork, CLI, and Web agent (no Browser MV3 extension or Office Add-in) | ◐ Partial Web workspace, Slack, and IDE integration | ◐ Partial Web console, CLI, and IDE bridge | ◐ Partial IDE, CLI, GitHub Web, Mobile | ◐ Partial IDE, GitLab Web UI, CLI |
Every metric below is reproduced continuously by our evaluation harness and documented with workload corpus details in our Benchmark Methodology Guide.
Complete 8-service sovereign stack (Keycloak 683 MiB, PostgreSQL 18 158 MiB, Gitea 93 MiB, SeaweedFS 78 MiB, Control Plane 29 MiB, OpenBao 26 MiB, Edge + OPA 25 MiB) verified at idle.
Methodology: Appliance Memory & Sizing Suite →Blocks test file deletion, .skip/.only injection, assertion removal, and snapshot tampering outside the agent sandbox.
Methodology: Foundry Verification Gate Suite →Deterministic changed-surface AST mutations verify that new tests genuinely fail when production logic is broken.
Methodology: AST Mutation Verification Suite →Hybrid Tree-sitter AST + SCIP/LSP call graph + pgvector semantic embeddings fused with Reciprocal Rank Fusion (3.5x fewer tokens than raw grep).
Methodology: Hybrid Code Intelligence Suite →Continuous Shadow Merge dry-runs across parallel git worktrees detect overlapping AST symbols in p95 = 0.01 ms.
Methodology: Multi-Worktree Radar Suite →Shadow model replay and deterministic prompt prefix ordering maximize cache hit rates without quality regression.
Methodology: Shadow FinOps Evaluation Suite →Mindaeon v0.1.0 is our Public Preview release. Every release candidate is gated against 289 automated capability contracts across our CLI runtime, Desktop/IDE surfaces, and multi-tenant Control Plane.
Covers Tree-sitter + SCIP indexing, parallel git-worktree execution, MCP client/server bridging, deterministic patch application, and offline air-gap mode.
Verifies real-time session handoff, diff review, DOM/network capture in the MV3 browser extension, VS Code diagnostics, and Office 365 document workflows.
Verifies PostgreSQL 18 RLS tenant isolation, OpenBao Transit signing, OPA Rego policy evaluation, 8-stage Foundry DAG execution, and 1,091 MiB appliance sizing.
Self-serve cloud trial workspaces include 5 seats, $5.00 in usage credits, dedicated PostgreSQL Row-Level Security, and immediate access to all 6 client surfaces.