Trail-Code README

The project's README from github.com/trailhq/Trail-Code. As one markdown file: README.md.

Graft — open-source context layer for large codebases

Turbocharge Claude Code, Cursor, Codex, Gemini & every coding agent: faster, cheaper, with contextual understanding specific to your codebase.

trailhq/Graft | Trendshift

Up to 4× cheaper and 3× faster, with better or no loss of correctness.

Metric Cold Claude Code Claude Code with graft
Tool-call reduction Baseline +46%
Token savings Baseline +42%
Time savings Baseline +60%
Correctness 54% 66% (+12 pts)

This works beyond code too.
A living skill file that learns from every task and gets sharper the more your team works.

Try Trail Brain

Side-by-side comparison of a coding agent working with and without graft

Contents

Quick start

npm install -g @nanonets/graft   # install the CLI, once
graft init                       # build the graph + wire it into Claude Code

That is the whole setup. graft init asks which of your coding agents to wire up, builds graft/ from your code, and drops a statusline and hooks into .claude/, so from the next session on Graft rides along in Claude Code: it pulls the matching nodes into each prompt and rebuilds the graph in the background after every turn. No daemon, no re-indexing to remember, nothing to run or maintain by default — the graph is just files.

Nothing is written until you pick. Run graft init --dry-run to see every file it would touch first, or graft init --agents claude to skip the prompt and wire Claude Code alone.

graft build adds graft/ to your .gitignore automatically — the graph is a local, regenerable cache (like node_modules), not something you commit. What you share is the wiring init dropped into .claude/; each teammate runs graft build to generate their own graph:

git add .claude && git commit -m "wire in graft"

Prefer not to install globally? npx @nanonets/graft init works the same way.

Two commands — npm install and graft init — then Graft rides along in a Claude Code session, statusline synced

The problem

Every task, your coding agent starts blind. Before it changes anything, it re-explores the repo: grep a term, open a file, follow an import, back out, try again. It is rebuilding a picture of a codebase it mapped an hour ago and threw away. That rediscovery burns most of a run's tool calls, tokens, and latency, and it is pure overhead:

Humans onboard to a codebase once. Agents onboard every single time.

A no-map agent's exploration trail wandering file to file before it finds what it needs

What Graft does

Graft builds that understanding once and writes it into your repo as a folder of linked markdown files, one node per system, API, or concept.

The same task, 'fix the auth bug', run two ways. A cold Claude Code session re-reads the repo and wanders file to file; Claude Code + graft loads its map once and rides the hooks to one clean pass. With Graft: 46% fewer tool calls, 42% fewer tokens, 60% less time, +22% more SWE-bench instances resolved.

Benchmark

An agent that reads the graph should be cheaper and faster without getting more answers wrong. That's the whole claim, so we measured it instead of asserting it.

The harness ran three variants of the same Claude Sonnet 5 agent with the same file tools: cold (explores from zero), Graft (a graft ask --source bundle pushed up front), and pull (graft_find_code/graft_file_api tools, nothing injected — context paid for only when asked). An Opus 4.8 judge scored correctness with a required-keyword floor, so a fast-but-wrong answer couldn't win by being fast. Cost is cache-aware: reads ≈0.1×, writes 1.25×, the billing model agents actually run under.

162 runs, two repos (graft itself and a real Node/Express auth service), 3 trials each, tasks split between single-file and multi-file questions.

Metric (mean/task) Cold Claude Code Claude Code with graft
Cost savings ($) 0.0429 0.0292 (+32%)
Token savings 8,070 4,650 (+42%)
Tool-call savings 4.2 2.3 (+46%)
Latency savings (s) 39.8 15.8 (+60%)
Correctness 93% 93% (equal)

Graft never answered worse than cold, on any corpus. The pull variant gave up most of that speed for something bigger: correctness jumped to 98%, +5 points over cold, the strongest single result in the sweep. Push when speed is what you need; pull when being right matters more.

SWE-bench Verified

The sweep above is our harness measuring our mechanism. So we ran the industry-standard one too — SWE-bench Verified, real GitHub issues from real repos, graded by the official swebench harness. No judge model, no similarity score: your patch is applied, the maintainers' own tests are run, and you either flip the failing test without breaking the passing ones or you don't.

50 instances, same model on both arms — Claude Sonnet 5 — same Docker images, same turn limits. The only difference is whether graft is wired in.

Correctness & efficiency Cold Claude Code Claude Code with graft Improvement
Correctness 27 / 50 (54%) 33 / 50 (66%) +12 pts
Token savings 142.0M 109.4M +23%
Cost savings $52.34 $42.43 +19%
Tool-call savings 1,370 1,031 +25%
API-request savings 2,455 1,875 +24%
Wall-clock savings 13,094s 8,922s +32%

graft resolved 33 of 50 instances against Cold Claude Code's 27 — and got there with 25% fewer tool calls, 23% fewer tokens, and 32% less wall-clock time. Every correctness win has the same shape: the baseline patches one file and misses its siblings. On django-11532 it patched 1 of the 5 files the fix requires and broke 18 previously-passing tests, twice over. On django-16263 it patched 1 of 4 and scored 102 / 103. graft found the rest — and on django-16263 did it in half the tokens and half the time.

Two harnesses, two claims: the controlled sweep says graft is cheaper and faster, SWE-bench says it's also more correct.

Correctness over all instances; tokens, cost and calls over the instances both arms resolved, for a like-for-like comparison. Official SWE-bench Verified images and official swebench 4.1.0 grader, native x86_64.

How the graph gets built

Graft builds the graph in two passes, both powered by a language model:

  1. Read each file. Every source file is summarized once into a short description of what it does.
  2. Group into nodes. Those summaries are grouped into a curated set of nodes (subsystems, key files, and concepts) with typed links between them. Graft chooses the right level of detail for you instead of making one node per file, so a big repo becomes a few dozen readable nodes.
flowchart LR
    S[Source files] --> T["Tier 1 — tree-sitter<br/>no model, no key"]
    S --> P1["Pass 1 — LLM summarizes<br/>each file (--deep)"]
    T --> W["graft/.graph/wiring.json<br/>per-symbol code graph"]
    P1 --> P2["Pass 2 — group into nodes<br/>+ typed links"]
    P2 --> N["graft/*.md<br/>markdown node graph"]

Every pass is cached by content hash — the LLM ones and the tree-sitter parse alike. Re-running only touches the files that changed, so the second build is fast and cheap (on this repo, 124 files: 0.74s cold, 0.18s after one edited file, 0.18s with nothing changed). graft build --no-reuse forces a cold re-parse.

That cheapness is what lets every query refresh the graph before it answers. A retrieval call stats the tree against the last build's fingerprint (~3ms), and rebuilds only if something moved — so ask/grep/callers/skeleton/map describe the code as it is right now, including edits that are unsaved to git: uncommitted, unstaged, or staged all look the same to graft. Git determines the visible file set; freshness compares the working-tree bytes rather than commit or index state. The refresh is structural and $0; it never calls the LLM. Turn it off per-command with --no-refresh, or everywhere with GRAFT_NO_REFRESH=1.

Alongside the markdown graph, graft build builds graft/.graph/wiring.json — a per-symbol code graph — plus a per-file wiring card mirroring your source tree. Tier 1 is pure tree-sitter (every function, class, and call edge; deterministic, no model, no network), which is why plain graft build needs no key. The --deep pass adds a one-line summary and a crux excerpt per symbol, cached by body hash.

Supported languages

Graft parses with tree-sitter at two levels of fidelity, plus an optional compiler-grade layer — all $0 and deterministic (no model, no key):

Twenty-three languages in total. A file whose language isn't listed is skipped, not indexed. Adding a broad-tier language is a small contribution — see CREDITS.md for the folks who added the current set.

What's in a node

A node is a single markdown file. Most code maps stop at an address: this thing lives in that file, on that line. That tells an agent where to look, not what it will find, so it still has to open the source and read. A Graft node holds the meaning inline, so the agent learns what it needs up front and opens the file only when it wants more.

Each node holds:

Part What it holds
Summary A plain-English explanation of what the code does, written by the model and cached. It is there whether or not the code was ever documented, and it is regenerated when the source changes.
Crux The handful of lines that actually carry the logic: the guard, the skip condition, the state change. Lifted straight from the source and stored inline, so the agent sees how it works, not just what.
Sources The exact files the node is built from, each tracked by a content hash, so Graft can tell precisely when a node has gone stale.
Links Typed connections to other nodes (depends_on, part_of, uses, implements, produces), written as [[wikilinks]] your agent can follow.
Notes Anything you write below the generated block. It is preserved across regenerations, so your own context is never overwritten.

That is three depths in one file: the summary says what the code does, the crux shows how, and the sources point to the rest if the agent needs it. A plain index makes it read a whole file to learn one thing. A Graft node hands it the answer inline, and the follow-up read often never happens.

The crux is stored as the code itself, not as a line range, on purpose. Line numbers drift whenever unrelated code above them shifts, but the lines that matter do not. Keeping the text, not the numbers, means the crux stays correct even as the file around it moves.

Summary, sources, links, and notes ship today in markdown nodes. The crux ships per-symbol in the code graph (graft build --deep); inlining it into markdown nodes is next.

What runs where

See .env.example for the full list of settings (model, base URL, graph directory).

Agent integration

One command wires Graft into the coding agents you use:

npx @nanonets/graft init
# detects your agents and writes each one's native instruction file;
# Claude Code additionally gets the live statusline + hooks below

On a terminal, init shows you every agent it knows about — flagging the ones it detected (via their config directories) and listing the exact files each would write — and wires only the ones you select. Claude Code is pre-selected; nothing else is. Selected agents get a marker-fenced Graft section in their shared instruction file — AGENTS.md (Codex, OpenCode and other CLIs that read it), GEMINI.md, .github/copilot-instructions.md — or a wholly-owned rule/skill file for the agents that use one: `.claud

Trail-Code (trailhq/Trail-Code)

Searching Database…

10.3K ·GitHub
Trail-Code
Что это такое?Graft превращает вашу кодовую базу в связанные markdown-файлы — по одному узлу на каждую систему, API или концепцию — и подсовывает подходящие узлы coding-агентам вроде Claude Code, Cursor, Codex и Gemini, чтобы им не приходилось заново исследовать репозиторий с каждой новой задачей: они работают быстрее, дешевле и точнее.
Graft — это облачный сервис, или всё хранится в моём репозитории?Это CLI плюс MCP-сервер, а сам граф — просто папка с markdown-файлами (graft/) внутри вашего репозитория: локальный регенерируемый кэш вроде node_modules. Ни демона, ни базы данных, и по умолчанию нечего запускать или поддерживать.
Что мне нужно сделать, чтобы мой coding-агент начал его использовать?Запустите 'graft init' — он спросит, какие агенты подключить, создаст graft/ и пропишет statusline и хуки в .claude/, так что граф подхватится уже со следующей сессии. Глобальная установка не нужна ('npx @nanonets/graft init'), а 'graft init --dry-run' заранее покажет каждый файл, который будет затронут.
Мне нужно использовать свой API-ключ?Нужен Node 20+. Структурный граф (graft build) построен целиком на tree-sitter и никогда не обращается к модели, поэтому ключ ему не требуется; только простые текстовые описания генерирует LLM — от любого выбранного вами провайдера: OpenAI, Anthropic, OpenRouter, Fireworks, Groq, прокси LiteLLM или локальная модель — под вашим собственным ключом.
Какие языки программирования он умеет разбирать?23 языка. Полная точность (с учётом областей видимости, разрешение вызовов и импортов между файлами) для TypeScript/JavaScript, включая JSX/TSX, Python, Go, Java, Kotlin, PHP, Swift и R; для Rust, C, C++, C#, Ruby, Scala и других — символы и связи вызовов. Опциональный проход LSP (rust-analyzer, clangd, gopls, pyright, tsserver) добавляет точные связи.
Если команда работает над общим репозиторием, каждому нужно строить свой граф?Да — graft/ это локальный кэш отдельного разработчика, и он автоматически попадает в gitignore. В репозиторий коммитится только небольшая обвязка, которую создал init (.claude/, AGENTS.md, конфиг MCP); каждый коллега один раз запускает graft build, чтобы построить свой граф.
Граф устареет, пока я правлю код?Нет — перед каждым запросом граф сверяется с вашим рабочим деревом, и это чисто структурная, бесплатная операция (~3 мс, если ничего не менялось); она видит незакоммиченные, нестейдженные и застейдженные правки одинаково. Сборки кэшируются по хэшу содержимого, так что повторный запуск затрагивает только изменённые файлы; graft check — локальный индикатор свежести.
Что именно находится внутри узла графа?Один markdown-файл: написанное моделью понятное описание того, что делает код, «ключевые» строки, в которых сосредоточена логика, точный список исходных файлов, отслеживаемых по хэшу содержимого, и типизированные [[вики-ссылки]] на другие узлы (depends-on, part-of, uses). Всё, что вы напишете ниже сгенерированного блока, сохраняется при повторной генерации.
Насколько реально быстрее и дешевле становятся агенты?По бенчмарку проекта из 162 запусков: на 42% меньше токенов, на 46% меньше вызовов инструментов, на 60% ниже задержка при той же корректности. На SWE-bench Verified (50 задач, тот же Claude Sonnet 5) он решил 33 задачи против 27 у холодного Claude Code — с на 23% меньшим числом токенов и на 32% меньшим временем выполнения.
Что с ним можно делать, кроме подключения Claude Code?Он работает как MCP-сервер для любого агента с поддержкой MCP. В CLI есть graft ask, grep, callers, skeleton и map для поиска и ориентации, graft viz для визуализации графа и graft build --deep для описаний по каждому символу. Режим push (контекст подаётся заранее) даёт скорость; режим pull в их замерах обменивал скорость на 98% корректности.
What are similar projects?graphicle, axiomcodegraph, graphify, GitNexus, headroom, claude-context, code-review-graph, jevmem
Which topics is it in?context-management, code-search, tree-sitter, code-analysis, knowledge-graph, context-engineering, ai-coding, visualization, monorepo, cursor, gemini, anthropic
Creator
trailhq
Size
69.9 MB
Stars
10.3K
Updated
CommitOct 15
Age
3mo
Commits · 90 Days