Struct IndexConfig 

pub struct IndexConfig {

    pub enabled: bool,
    pub search_enabled: bool,
    pub watch: bool,
    pub max_chunks: usize,
    pub score_threshold: f32,
    pub budget_ratio: f32,
    pub repo_map_tokens: usize,
    pub repo_map_ttl_secs: u64,
    pub mcp_enabled: bool,
    pub workspace_root: Option<PathBuf>,
    pub concurrency: usize,
    pub batch_size: usize,
    pub memory_batch_size: usize,
    pub max_file_bytes: usize,
    pub embed_provider: Option<String>,
    pub embed_concurrency: usize,
}

Code indexing and repo-map configuration, nested under [index] in TOML.

When enabled = true, the agent indexes source files into Qdrant for semantic code search. The repo map is injected into the system prompt or served via IndexMcpServer tool calls when mcp_enabled = true.

Example (TOML)

[index]
enabled = true
watch = false
max_chunks = 12
score_threshold = 0.25

Fields

enabled: bool

Enable code indexing. Default: false.

search_enabled: bool

Enable semantic code search tool. Default: true (no-op when enabled = false).

watch: bool

max_chunks: usize

score_threshold: f32

budget_ratio: f32

repo_map_tokens: usize

repo_map_ttl_secs: u64

mcp_enabled: bool

Enable IndexMcpServer tools (symbol_definition, find_text_references, call_graph, module_summary). When true, static repo-map injection is skipped and the LLM uses on-demand tool calls instead.

workspace_root: Option<PathBuf>

Root directory to index. When None, falls back to the current working directory at startup. Relative paths are resolved relative to the process working directory.

concurrency: usize

Number of files to process concurrently during initial indexing. Default: 4.

batch_size: usize

Maximum number of new chunks to batch into a single Qdrant upsert per file. Default: 32.

memory_batch_size: usize

Number of files to process per memory batch during initial indexing. After each batch the stream is dropped and the executor yields to allow the allocator to reclaim pages. Default: 32.

max_file_bytes: usize

Maximum file size in bytes to index. Files larger than this are skipped. Protects against large generated files (e.g. lock files, minified JS). Default: 512 KiB.

embed_provider: Option<String>

Name of a [[llm.providers]] entry to use exclusively for embedding calls during indexing. A dedicated provider prevents the indexer from contending with the guardrail at the API server level (rate limits, Ollama single-model lock). When unset or empty, falls back to the main agent provider.

embed_concurrency: usize

Maximum parallel embed_batch calls during indexing (default: 2 to stay within provider TPM limits).

Trait Implementations

impl Debug for IndexConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for IndexConfig

fn default() -> IndexConfig

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for IndexConfig

fn deserialize<__D>( __deserializer: __D, ) -> Result<IndexConfig, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for IndexConfig

fn serialize<__S>( &self, __serializer: __S, ) -> Result<<__S as Serializer>::Ok, <__S as Serializer>::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.