feat(server): split camelCase identifiers in embeddings and stamp stored vectors

Cargo.toml

@@ -60,7 +60,7 @@ members = [
 ]
 
 [workspace.package]
-version = "0.20.1"
+version = "0.21.0"
 edition = "2021"
 license = "Apache-2.0"
 repository = "https://github.com/codegraph-ai/codegraph"

README.md

@@ -30,7 +30,7 @@ The server indexes the current working directory automatically.
 Install the VSIX:
 
 ```bash
-code --install-extension codegraph-0.20.1.vsix
+code --install-extension codegraph-0.21.0.vsix
 ```
 
 One VSIX serves every platform.
@@ -85,6 +85,23 @@ one tool and exits without the MCP stdio handshake — ideal for scripting.
 
 ---
 
+## Upgrading to 0.21
+
+Each project re-embeds once, in the background, the first time 0.21 opens it.
+Stored vectors now record the model and settings that built them, and an index from 0.20.1 or earlier records none.
+Keyword search keeps working while it runs.
+
+Restart any `--watch` daemon after upgrading.
+A daemon left running keeps writing vectors that 0.21 will not load, and sessions attached to it run without semantic search until it restarts.
+
+Identifiers whose words run together, like `getUserById`, are now also embedded in word-split form.
+This mostly helps natural-language search in camelCase languages such as TypeScript, Java and C#.
+Pass `--split-identifiers=false` to keep the old behaviour.
+
+`--full-body-embedding=false` now takes effect; in 0.20.1 the flag was always on.
+
+---
+
 ## Configuration
 
 ### MCP Server flags
@@ -94,12 +111,19 @@ one tool and exits without the MCP stdio handshake — ideal for scripting.
 | `--workspace <path>` | current dir | Directories to index (repeatable for multi-project) |
 | `--exclude <dir>` | — | Directories to skip (repeatable) |
 | `--embedding-model <model>` | `bge-small` | `bge-small` (384d, fast), `jina-code-v2` (768d, 6× slower), `granite-97m` (384d, 32K ctx, ~3× slower), or `static` (model2vec, 256d — ~100× faster indexing, no ONNX; needs a local model dir, see below) |
-| `--full-body-embedding` | `true` | Embed full function body (~50 lines) for better semantic search and duplicate detection |
+| `--full-body-embedding` | `true` | Embed full function body (~50 lines) for better semantic search and duplicate detection. Takes a value: `--full-body-embedding=false` turns it off |
+| `--split-identifiers` | `true` | Also embed the word-split form of identifiers whose words are run together, so `getUserById` embeds as "get user by id" too. Names already separated by `_` or `-` are left alone, since they tokenise into the same words; a leading or trailing delimiter separates nothing, so `_handleClick` is split like `handleClick`. Takes a value: `--split-identifiers=false` embeds raw names, as releases up to 0.20.1 did |
 | `--max-files <n>` | 5000 | Maximum files to index |
 | `--profile <name>` | `all` | Filter the exposed MCP tool surface to a named subset (see below) |
 | `--graph-only` | off | Skip embedding generation — build the graph and serve structural tools only. No ONNX model load, 10-50× faster indexing. Semantic search and memory tools unavailable. For CI / one-shot graph queries. |
 | `--run-tool <name>` | — | One-shot mode: index, run a single tool, print its result, exit. No MCP handshake. Pair with `--tool-args '<json>'`. |
 
+`--split-identifiers` and `--full-body-embedding` both change the text every symbol is embedded from, and vectors built from different text cannot be ranked against each other.
+A project's stored vectors are therefore stamped with the settings that built them - `--embedding-model` included, since models differ in dimension - and are ignored by any process configured differently.
+Changing any of the three re-embeds in the background rather than requiring a manual reindex.
+Run `--watch` with the same flags as the sessions that read the project, so both sides share one set instead of re-embedding over each other.
+For what this means when upgrading from 0.20.1 or earlier, see [Upgrading to 0.21](#upgrading-to-021).
+
 #### `--embedding-model static` — model2vec fast indexing
 
 Static (model2vec) embeddings replace the ONNX transformer with a token→vector

crates/codegraph-memory/examples/embed_eval.rs

@@ -168,7 +168,15 @@ fn metrics(ranks: &[usize]) -> Scores {
 
 /// Returns (semantic-only, hybrid 0.4*BM25 + 0.6*cosine) scores.
 fn evaluate(engine: &VectorEngine, syms: &[Sym]) -> (Scores, Scores) {
-    let split = std::env::var("CODEGRAPH_SPLIT_IDS").is_ok();
+    // Presence is not the question - `CODEGRAPH_SPLIT_IDS=0` must mean off.
+    // `is_ok()` made setting it to 0 turn the feature ON, which silently turned
+    // an A/B run into two identical arms. `=1` is the one spelling this harness
+    // documents, so it is the one spelling it accepts.
+    //
+    // Note this forces splitting on every identifier, unlike the engine, which
+    // applies it only to delimiter-free names: the point of this harness is to
+    // measure the lever in isolation.
+    let split = std::env::var("CODEGRAPH_SPLIT_IDS").as_deref() == Ok("1");
     let sym_texts: Vec<String> = syms
         .iter()
         .map(|s| {