Skip to content

Latest commit

 

History

History
99 lines (63 loc) · 3.19 KB

CONTRIBUTING.md

File metadata and controls

99 lines (63 loc) · 3.19 KB

Contributing to CodeGraph

Thanks for helping improve CodeGraph. This repo is a Rust workspace with multiple crates under crates/, plus SurrealDB schemas under schema/.

Quick start

  1. Install Rust (via rustup) and any platform prerequisites from docs/INSTALLATION_GUIDE.md.
  2. Build and run a fast hygiene pass:
cargo check --workspace
  1. Run tests (start with the crate you touched):
cargo test -p codegraph-mcp
  1. Format before you send a PR:
cargo fmt

What to change where

Indexing pipeline

  • Orchestration and analyzer phases: crates/codegraph-mcp/src/indexer.rs
  • Analyzer implementations: crates/codegraph-mcp/src/analyzers/
  • Parser and AST extraction: crates/codegraph-parser/src/

Embeddings and reranking

  • Provider implementations: crates/codegraph-vector/src/
  • Runtime config loader (TOML + .env + env overrides): crates/codegraph-core/src/config_manager.rs
  • Provider-specific config structs: crates/codegraph-core/src/config_manager.rs and crates/codegraph-core/src/rerank_config.rs

Built-in agent (agentic MCP tools)

  • MCP server and tool entrypoints: crates/codegraph-mcp-server/src/official_server.rs
  • Prompt tier selection: crates/codegraph-mcp-server/src/prompt_selector.rs
  • Tier prompts per analysis type: crates/codegraph-mcp-server/src/*_prompts.rs
  • LLM providers: crates/codegraph-ai/src/

Adding a new language

CodeGraph language support typically requires three layers:

  1. Core language enum: add or confirm the language exists in crates/codegraph-core/src/types.rs
  2. Tree-sitter registration: add the grammar dependency and register extensions in crates/codegraph-parser/src/language.rs
  3. Extraction: implement node/edge extraction for the new language in the parser pipeline (follow the existing extractor patterns in crates/codegraph-parser/src/)

After you add a language:

  • Update docs/SUPPORTED_LANGUAGES.md
  • Add or update tests that validate language registration and basic parsing

Adding an analyzer

Analyzers run during codegraph index to enrich the initial AST graph.

Guidelines:

  • Prefer small, composable analyzers that add clearly attributable nodes/edges.
  • Ensure analyzer output is deterministic and scoped by project_id.
  • Emit provenance metadata (analyzer name, confidence) so later tools can explain where facts came from.

If your analyzer depends on external tools (e.g. language servers), add the tool requirement in:

  • crates/codegraph-mcp/src/analyzers/mod.rs

Schema changes (SurrealDB)

Schema files live in schema/:

  • schema/codegraph.surql (default schema)
  • schema/codegraph_graph_experimental.surql (experimental graph-oriented schema)

If you change a schema:

  • Update any schema-level tests that validate required functions/indexes.
  • Keep SurrealQL compatible with current SurrealDB parsing rules.

Documentation updates

If your change affects configuration, providers, or language support, update:

  • docs/AI_PROVIDERS.md
  • docs/AGENT_PROMPT_TIERS.md
  • docs/SUPPORTED_LANGUAGES.md

PR checklist

  • cargo test passes for touched crates
  • cargo fmt clean
  • No new secrets committed
  • Docs updated for user-facing behavior changes