whereismyai
arcKode
arcKode: Local-first AI coding agent that indexes codebases using Tree-sitter graphs to plan and apply edits through Ollama.
arcKode is a local-first AI coding agent that operates within compact VRAM constraints. It parses repositories with Tree-sitter across 11 languages to build a live symbol graph containing functions, classes, imports, and call edges. This graph scopes the exact code context sent to the model, preventing hallucinated file paths. Edits are generated as ordered sequences across files and applied as targeted patches against anchored positions in source files. The tool runs locally via Ollama without cloud calls, recommends qwen2.5-coder:7b, and exports codebase.json maps.
arcKode at a glance
Key features
- Zero external dependencies via local Ollama models
- Semantic code parsing using Tree-sitter across 11 languages
- Token-aware context packing with reciprocal-rank fusion
- Two-mode operation for single-shot edits or multi-step refactors
- Built-in safety snapshots, anchored edits, and path containment checks
- Incremental indexing powered by SHA-256 change detection
- Exportable codebase.json maps for other coding agents
Use it for
- Execute surgical single-file code edits locally
- Run planned multi-step refactors across multiple files
- Generate high-level codebase maps for other agents
- Inspect symbol-level indexing and search project files via FTS5
Companions
Tools that articles about arcKode recommend using with it, each with its job.
- Ollama — Built to work with Ollama.
- Open WebUI — Run AI on your own terms with complete data privacy and self-hosting control.
- AnythingLLM — Runs fully on-device with no accounts, API keys, or token limits required.
- LangChain — Framework for custom agents with memory and tools on local models.
- Chroma — Local vector database for embeddings and semantic search.
- Cline — Open source, any model, with approval on every step
Alternatives to arcKode
- CodeGraph — It eliminates file crawling, reducing AI agent tool calls by 88% and cutting costs.
- Code Review Graph — Provides a median 63x context reduction while keeping all code processing local.
- OpenCode — Works with 75+ model providers, local models, and existing subscriptions
- SmallCode — Runs fully local with budget-managed context and a forgiving parser for small models.
- Strata — Runs large AI models locally on consumer gaming PCs without server reliance.
- Codebase Memory MCP — High-speed indexing and persistent knowledge graphs for AI coding agents.
- Archon — Orchestrate multiple AI agents with isolated workflows via a single command layer.
- Kolega Labs — It runs locally, supports multiple providers, and journals every phase for audits.
- jcode — Optimized for massive parallelism with low memory usage per session.
- Kilo Code — Open-source agent with any model and no markup on inference
- Prokop — Agents maintain memory and skills across projects and sessions.
- Enzyme — It indexes markdown notes locally so AI coding agents can cite them.
All arcKode alternatives