syntext

A faster grep for agentic AI. ~20X faster than ripgrep when indexed.

Hybrid code search index for agent workflows, built in Rust. Indexes repositories using sparse n-grams, then narrows to a small candidate set before verification. Drop-in replacement for rg in AI agent loops where grep is called repeatedly and in parallel.

Status: stable (v1.0).

Installation

Quick install (macOS and Linux)

curl -fsSL https://raw.githubusercontent.com/whit3rabbit/syntext/main/install.sh | sh

Installs st to /usr/local/bin. On macOS, uses Homebrew cask if brew is available. On Debian/Ubuntu (x86_64), installs the .deb package. All other Linux targets get the raw binary. Checksums are verified against SHA256SUMS from the release.

Override defaults with environment variables:

INSTALL_DIR=~/.local/bin SYNTEXT_VERSION=1.0.1 \
  curl -fsSL https://raw.githubusercontent.com/whit3rabbit/syntext/main/install.sh | sh

brew tap whit3rabbit/tap
brew install --cask whit3rabbit/tap/syntext

VERSION=1.0.1

# Debian/Ubuntu (x86_64)
curl -L "https://github.com/whit3rabbit/syntext/releases/download/v${VERSION}/syntext_${VERSION}_amd64.deb" \
  -o "syntext_${VERSION}_amd64.deb"
sudo dpkg -i "syntext_${VERSION}_amd64.deb"

# Any Linux (x86_64 or arm64)
ARCH=amd64   # or arm64
curl -L "https://github.com/whit3rabbit/syntext/releases/download/v${VERSION}/st-${VERSION}-linux-${ARCH}" -o st
chmod +x st && sudo mv st /usr/local/bin/

iwr -useb https://raw.githubusercontent.com/whit3rabbit/syntext/main/install.ps1 | iex

Installs st.exe to %LOCALAPPDATA%\syntext and adds it to the user PATH. Restart your terminal after install.

To pin a version or run from a saved script:

powershell -ExecutionPolicy Bypass -File install.ps1

Prebuilt WASM packages are available on the releases page as syntext-wasm-<version>.tar.gz. To build from source:

cargo install wasm-pack
wasm-pack build --target bundler -- --features wasm --no-default-features
# output: pkg/  (JS glue + .wasm + TypeScript types)

Other targets: --target nodejs, --target web.

From source

cargo install syntext

Benchmarks

Search latency across five real-world repositories (v1.0, macOS, Apple Silicon).

Repo	`st` avg	`rg` avg	`grep` avg	Speedup vs `rg`
React	`20.7 ms`	`112.9 ms`	`314.3 ms`	`5.5x`
Rust compiler	`99.9 ms`	`2183.2 ms`	`2412.8 ms`	`21.9x`
TypeScript	`111.9 ms`	`3093.8 ms`	`3171.8 ms`	`27.7x`
Node.js	`69.5 ms`	`1492.6 ms`	`3186.4 ms`	`21.5x`
Linux kernel	`154.5 ms`	`3681.3 ms`	n/a	`23.8x`

Average speedup across five presets: 20.1x versus rg. Search time excludes index build time.

See docs/BENCHMARKS.md for methodology, index build times, query discipline, and historical runs.

Usage

# Build the index (run once per repo, then only after large changes)
# Index is stored in .syntext/ at the repo root (nearest .git ancestor).
# Not run automatically -- you must run this before the first search.
st index
st index --stats                    # show file count and index size after build

# Override where the index is stored or which root to index
st --repo-root /path/to/repo index
st --index-dir /tmp/my-index index

# After editing files, sync the index incrementally (faster than full rebuild)
st update

# Search the whole repo (index must exist)
st "fn parse_query"                 # regex
st -F "parse_query("                # literal (metacharacters stay literal)
st -i "parsequery"                  # case-insensitive
st -x "TODO"                        # whole-line match
st -n "impl.*Iterator"              # force line numbers

# Restrict search scope with positional paths
st "needle" src/                    # search one directory
st "needle" src/lib.rs              # search one file
st "needle" src/lib.rs tests/       # search multiple files/directories

# Additional filters and output modes
st -t rs "impl.*Iterator"           # restrict to Rust files
st -g "src/" "TODO"                 # restrict by glob
st -c "parse_query" src/lib.rs      # count matches in one file
st -l "parse_query"                 # print matching file paths
st --json "TODO"                    # NDJSON output for tooling

# Status
st status

Notes:

Search is the default command, there is no st search subcommand.
Like ripgrep, file names are shown by default when searching a directory, the whole repo, or multiple positional paths.
Like ripgrep, line numbers are off by default when stdout is not a TTY. Use -n to force them on.

Agent configuration

To tell an AI agent to use st instead of rg or grep, add the following to your CLAUDE.md, AGENTS.md, or equivalent agent instruction file. The key constraint: check for the index once, not on every search.

## Code search

Use `st` instead of `rg` or `grep` for all code searches. `st` is a
drop-in replacement for ripgrep: same flags, identical output, but searches
a pre-built index and is significantly faster on repeated queries.

Before the first search in a session, check whether the index exists:

    test -d .syntext || st index

Do not check for the index on every search. Once built, assume it is valid
for the session. If files change mid-task, run `st update` to sync
incrementally instead of rebuilding.

Common usage (same flags as rg):

    st "pattern"              # regex search
    st -F "literal string"    # fixed string, no regex interpretation
    st -i "pattern"           # case-insensitive
    st -t rs "pattern"        # restrict to file type (e.g. rs, py, ts)
    st -l "pattern"           # list matching files only
    st -n "pattern"           # include line numbers
    st "pattern" src/         # restrict to a directory
    st --json "pattern"       # machine-readable NDJSON output

Architecture

Query -> Router -> [Literal | Indexed Regex | Full Scan]
                        |
                   Gram extraction
                        |
                   Posting list intersection (smallest-first)
                        |
                   Candidate file IDs
                        |
                   Verifier (memchr or regex against file content)
                        |
                   Results

Three index components:

Content index: sparse n-gram posting lists. Trigram augmentation ensures no false negatives for token-aligned queries.
Path index: Roaring bitmap component sets for path/type filtering.
Symbol index (optional): Tree-sitter extraction into SQLite.

Segments are immutable single-file mmap structures (SNTX format). Updates go through an in-memory overlay with atomic batch commit via ArcSwap.

See docs/ARCHITECTURE.md for the full quantitative analysis: selectivity math, index size estimates, posting list encoding tradeoffs.

WASM

The wasm Cargo feature compiles syntext to a fully in-memory index with no filesystem access. See the releases page for prebuilt syntext-wasm-<version>.tar.gz, or build from source:

wasm-pack build --target bundler -- --features wasm --no-default-features
# output: pkg/  (JS glue + .wasm + TypeScript types)

Project status

All phases complete (v1.0). Core st index && st "pattern" workflow validated against ripgrep. Symbol search available behind --features symbols.

Phase	Status	What it delivers
1. Setup	Complete	Cargo project, dependencies, module structure
2. Foundational	Complete	Weight table, tokenizer, posting lists, correctness harness
3. US5 -- Build	Complete	Full index build from scratch
4. US1 -- Search	Complete	Literal + regex search, ripgrep correctness validation
5. US2 -- Incremental	Complete	Overlay, batch commit, read-your-writes
6. US3 -- Path scoping	Complete	Path/type filters with Roaring bitmaps
7. US4 -- Symbols	Complete	Tree-sitter symbol extraction, SQLite storage
8. CLI	Complete	`st` binary with grep-compatible output
9. Polish	Complete	Bug fixes, security hardening, benchmarks, documentation

Known limitations

Crash recovery: Overlay state is lost on unclean shutdown. Run st update or st index after a crash.
Non-aligned substring coverage: ~16% false-negative rate for queries that don't align with token boundaries. Token-aligned queries (identifiers, keywords) have 0% false negatives.
Network filesystems: Index directory must be on local filesystem. NFS/SMB behavior is undefined.
Case-insensitive overhead: ~15-20% more candidates due to lowercase normalization. Correct results guaranteed by verifier.
\r-only line endings: Treated as a single line (matches ripgrep behavior).
Symbol search accuracy: Tier 3 (heuristic) results are approximate. Tree-sitter failures fall back silently.
One root per index: Each index covers exactly one --repo-root. There is no way to merge multiple directories into a single index. To search across two repos, build and query each index separately with --repo-root. st update requires a git repo; non-git directories must be re-indexed with st index.

Design documents

docs/ARCHITECTURE.md -- Quantitative analysis: selectivity math, index size estimates, posting list encoding, design tradeoffs
specs/001-hybrid-code-search-index/spec.md -- Feature specification with user stories and acceptance criteria
specs/001-hybrid-code-search-index/research.md -- 19-section architecture research covering every subsystem
specs/001-hybrid-code-search-index/data-model.md -- Entity definitions and relationships
specs/001-hybrid-code-search-index/contracts/ -- Library API, CLI, and segment format contracts
specs/001-hybrid-code-search-index/tasks.md -- Implementation plan with dependency graph

License

MIT

syntext 1.0.2