Frequently Asked Questions

Which coding agents does DAMS support?

OpenCode, Cline, Claude Code, Cursor, Continue.dev, Kilo Code, Antigravity CLI/GUI, and any tool that reads a Markdown steering file (e.g., DAMS.md, AGENTS.md, CBM.md). dams configure-agents . installs OS shims and steering rules for all of them.

Do I need a GPU?

A GPU accelerates local embeddings and reranking, but DAMS also supports cloud LLM providers (OpenAI-compatible) for Chunkhound Code Research. The knowledge graph builds from Tree-sitter AST alone (--code-only) with no LLM and no GPU.

Is my code sent to the cloud?

No, by default. Indexing, embeddings, reranking, and research run locally on Ollama. Cloud use is opt-in and limited to the LLM provider you configure for research synthesis.

What if Graphify needs an API key?

Full semantic extraction needs an LLM key, but --code-only mode uses Tree-sitter only and still delivers the 71.5× token reduction on subgraph queries. The OpenAI backend is patched to honor OPENAI_BASE_URL / GRAPHIFY_OPENAI_MODEL, so any OpenAI-compatible gateway (NanoGPT, LMStudio, OpenRouter, MiniMax) works.

How are tasks tracked?

Tasks are Markdown files in .dams/context/tasks/ plus a machine-readable _registry.yaml and a SQLite ledger. A [TASK-ID] in a commit message auto-marks the task complete via the post-commit hook.

Can multiple developers share context?

Yes. Steering configurations and graphify-out/ outputs are git-synchronized; dams sync-push / sync-pull handle multi-machine sync, and export-context / import-context support cross-project sharing.

How do I run DAMS?

The recommended runtime is the included .devcontainer (a single docker compose up). Inside it, dams ui starts both the FastAPI backend (8765) and Vite UI (5174).

What's the project status?

DAMS v2.3 is finalized — all core components and iterative tooling (Phases 1–8) are complete and production-ready.