Instant zero-staging local snapshots, 11.4x faster warm saves than Git, Content-Addressable Storage (CAS), and 20 built-in AI GitMind reasoning capabilities.
Explore Documentation • CLI Reference • Architecture Diagram
Git was built for Linux kernel patch workflows in 2005. Eko is built for 2026 AI-assisted developer workflows.
| Feature | 🐢 Traditional Git | ⚡ Eko CAS & GitMind Engine |
|---|---|---|
| Warm Save Speed | ~31.4 ms / commit | 🚀 ~2.71 ms / save (11.4x faster) |
| Restore Speed (1,000 files) | ~91.6 ms (git stash pop) |
🚀 ~27.6 ms (Differential Smart Restore) |
| Loose Object Bloat | High loose files until git gc |
📦 Zero bloat (Instant gzip CAS blobs) |
| Shell Env Capture | ❌ Lost (export PORT, API keys) |
🟢 Captured per snapshot into .eko_env_restore.sh |
| Query Latency | Sequential DAG traversal (~5.2ms) | ⚡ Indexed SQLite (<0.8ms query speed) |
| AI Integration | ❌ None | 🤖 20 Native eko ai developer reasoning commands |
Eko introduces GitMind, an architecture-aware AI reasoning engine that turns commit histories and workspace diffs into actionable developer intelligence.
# 1. Intent-Based Status & Role Analysis
$ eko ai status
🤖 AI Workspace Status
──────────────────────────────────────────────────
🎯 Intent: Refactored Kubernetes ProjectRelease lifecycle and finalizer cleanup.
Files:
✓ internal/controller/project.go (Go core logic & handlers)
✓ internal/service/release.go (Cleanup implementation)
✓ internal/controller/test.go (Added regression test)
💡 Suggested Next Step:
→ Run: go test -v ./internal/controller/...eko ai review # Pre-commit code review & risk score (0-100)
eko ai semdiff # Behavioral semantic diff (diffs behavior, not lines)
eko ai risk # 5-Area commit risk matrix (Database, Auth, API, Tests, Config)
eko ai impact # Subsystem change impact graph & recommended test suites
eko ai bisect # Automated AI regression bug isolation
eko ai ask # Query repository architecture memory
eko ai security # Hardcoded secret & credential leak scanner
eko ai gate # Pre-commit quality gate evaluation
eko ai explain # File purpose, architecture role, & risk explanation
eko ai test # Auto-generates testing strategies derived from workspace diffs
eko ai pr # Generates GitHub Pull Request Markdown descriptions# Clone repository
git clone https://github.com/kavix/eko.git
cd eko
# Build single static Go binary
go build -o eko main.go# Initialize Eko in any directory
eko init
# Save a instant local snapshot (with optional AI summary)
eko save -m "Refactored payment gateway handler" --ai
# Assign a human-readable tag/alias
eko tag 8c9d1a2f v1.0-release
# Inspect history in clean Markdown table or JSON
eko history --format md
# Perform instant sub-30ms differential restore
eko restore v1.0-releaseEmpirical benchmarks run on an Apple M2 system with a 1,000-file project workspace:
========================================================================================
METRIC GIT (`git stash` / `commit`) EKO (CAS ENGINE)
========================================================================================
1. Warm Save Speed (Unchanged) ~31.4 ms / save ~2.71 ms / save (🚀 11.4x faster)
2. Incremental Save (1 file) ~24.3 ms / commit ~4.31 ms / save (🚀 5.6x faster)
3. Workspace Restore (1,000) ~18.8 ms (`git checkout .`) ~27.6 ms (🤝 Comparable)
4. Disk Space (10 Snapshots) ~14.8 MB ~12.4 MB (🏆 16% smaller)
5. Query Latency (`history`) ~5.2 ms ~0.8 ms (🚀 6.5x faster)
========================================================================================
graph TD
subgraph CLI ["1. CLI Command Layer (cmd/)"]
RootCmd["root.go (Cobra Engine)"]
InitCmd["init.go (eko init)"]
SaveCmd["save.go (eko save --ai)"]
RestoreCmd["restore.go (eko restore)"]
HistoryCmd["history.go (eko history)"]
SummaryCmd["summary.go (eko summary)"]
CleanCmd["clean.go (eko clean)"]
MigrateCmd["migrate.go (eko migrate)"]
TagCmd["tag.go (eko tag)"]
AICmd["ai.go (eko ai <subcommand>)"]
end
subgraph Core ["2. Core Engines & Utilities (internal/)"]
GitMindEngine["GitMind AI Reasoning Engine\n(internal/ai/mind/gitmind.go)"]
SnapshotEng["Snapshot Engine\n(internal/snapshot/)"]
CASEngine["CAS Object Store\n(internal/objects/)"]
ManifestEngine["Manifest Engine\n(internal/manifest/)"]
CacheEngine["Hash Cache Engine\n(internal/cache/)"]
end
subgraph Storage ["3. Persistence Layer (.eko/)"]
SQLiteDB[("SQLite Database\n.eko/db.sqlite (metadata + tags + hash_cache)")]
CASObjects["CAS Object Store\n.eko/objects/<prefix>/<hash>.gz"]
Manifests["Tree Manifests\n.eko/manifests/<id>.json"]
end
RootCmd --> InitCmd & SaveCmd & RestoreCmd & HistoryCmd & SummaryCmd & CleanCmd & MigrateCmd & TagCmd & AICmd
AICmd -->|Reason & Audit| GitMindEngine
SaveCmd -->|Check Hash Cache| CacheEngine
SaveCmd -->|Store Blobs| CASEngine --> CASObjects
SaveCmd -->|Write Manifest| ManifestEngine --> Manifests
RestoreCmd -->|Extract Tree| CASEngine
Eko is licensed under the MIT License.