lidx Development Workflows

When to Use This Skill

• •Adding a new RPC method
• •Adding a new language extractor
• •Adding a database migration
• •Writing or running tests
• •Running the project locally

Common Workflows

Adding a New RPC Method

1. •

   Define params in src/rpc.rs:

   rust

   #[derive(Deserialize)]
   struct MyMethodParams { field: String, limit: Option<usize> }
   

2. •

   Add to METHOD_NAMES array (around line 766)

3. •

   Add MethodDoc in method_docs() with name, summary, param schema, examples

4. •

   Add dispatch arm in handle_method():

   rust

   "my_method" => {
       let params: MyMethodParams = serde_json::from_value(params_val)?;
       let result = json!({ "data": data, "next_hops": next_hops });
       Ok(result)
   }
   

5. •

   Add response type to src/model.rs if needed (derive Debug, Serialize, Clone)

6. •

   Add DB query to src/db/mod.rs if needed

Adding a Language Extractor

1. •

   Add tree-sitter dependency: cargo add tree-sitter-{lang}

2. •

   Create src/indexer/{lang}.rs:

   rust

   pub struct LangExtractor { parser: Parser }
   impl LangExtractor {
       pub fn new() -> Result<Self> { /* init parser */ }
       pub fn extract(&mut self, source: &str, module: &str) -> Result<ExtractedFile> { /* walk AST */ }
   }
   

3. •

   Register in src/indexer/mod.rs: add to LanguageKind enum, instantiate in Indexer::new_with_options, add match in index_file

4. •

   Map extensions in src/indexer/scan.rs: add to the extension match

5. •

   Write test: tests/{lang}_extract.rs with fixture in tests/fixtures/sample.{ext}

Adding a DB Migration

1. •Open src/db/migrations.rs
2. •Bump SCHEMA_VERSION (currently 11)
3. •Add version-gated block:
   rust

   if existing < 12 {
       if !has_column(conn, "table", "new_col")? {
           conn.execute("ALTER TABLE table ADD COLUMN new_col TEXT", [])?;
       }
   }
   

4. •Add index if the column will be queried:
   rust

   conn.execute("CREATE INDEX IF NOT EXISTS idx_table_col ON table(new_col)", [])?;
   

Writing Tests

• •Integration tests: tests/ directory, each file is a separate test binary
• •Fixtures: tests/fixtures/ -- real source files for extraction tests
• •Pattern: create temp dir, init Indexer, index fixtures, assert on Db query results
• •Run all: cargo test
• •Run specific: cargo test --test my_test_file
• •With output: cargo test -- --nocapture
• •Single test fn: cargo test --test my_test_file test_fn_name

Running the Project

# MCP server
cargo run --release -- mcp-serve --repo /absolute/path/to/repo

# JSONL RPC server
cargo run --release -- serve --repo /absolute/path/to/repo

# One-shot reindex
cargo run --release -- reindex --repo /absolute/path/to/repo

# Single RPC request
cargo run --release -- request --repo /path --method find_symbol --params '{"query":"Indexer"}'

# Test a method via pipe
echo '{"method":"repo_overview","params":{"summary":true}}' | cargo run --release -- serve --repo /path

Always use absolute paths -- ~ and relative paths are not expanded.

Pre-commit Checks

cargo fmt --check && cargo clippy && cargo test

Quick Reference

Add RPC method: params struct -> METHOD_NAMES -> MethodDoc -> dispatch arm -> model type -> DB query Add extractor: tree-sitter dep -> indexer/{lang}.rs -> register in mod.rs -> map extension in scan.rs -> test Add migration: bump SCHEMA_VERSION -> version-gated block in migrations.rs Run tests: cargo test or cargo test --test <name> Run server: cargo run --release -- mcp-serve --repo /absolute/path