LeanToken

Make every AI coding token go further

Local-first code intelligence for coding agents. Search code, inspect structure, read exact ranges, and explore Git history through a CLI and MCP server.

Language: English · 简体中文 · 日本語 · 한국어

Install · Why LeanToken · Tools · CLI · How it works · Docs

---

Measured token savings: In a controlled 60-run study, LeanToken used 20.1% fewer model input tokens than the agent's built-in tools with limited repository exploration, and 37.6% fewer than those tools with broad exploration. See exactly how it was measured in the measurement methodology.

Quick start

Add LeanToken to Claude Code, Cursor, OpenCode, Codex, Gemini CLI, or Antigravity:

npx leantoken setup

Setup behavior and safety

Current releases stop setup before writing when npx resolves a stale project-local or ancestor install, and point to npx leantoken@latest setup. Older releases that predate this check can be bootstrapped directly with that versioned command.

The setup wizard labels supported clients it detects, but leaves every client unselected so you choose exactly which coding agents receive LeanToken. Before writing anything, it shows the exact configuration paths and MCP launcher and asks for confirmation. An npx-based setup pins the exact LeanToken version that ran setup, so restarting a client cannot silently move to a newer release.

Global setup never stores the repository where setup happened. OpenCode gets a workspace-relative working directory; other supported clients launch LeanToken from the workspace cwd selected by the host. If a host instead starts it from the home directory or a filesystem root, LeanToken refuses to index that broad root by default.

Restart or reload the configured clients, then verify the connection and first retrieval from a repository:

npx leantoken doctor

Try a broad task such as: Find the code related to request cancellation before editing. LeanToken helps the agent start with leantoken.context, while its normal tools remain available for edits, builds, and tests.

Inspect LeanToken's observed repository-local token accounting:

npx leantoken savings

Local by default Source is indexed on your machine in a local database. LeanToken is a read-only discovery and retrieval layer.

Explicit token budgets Every response has an explicit token limit, so large files cannot take over the request.

Built for agent workflows Find files, search code, inspect structure, read exact ranges, trace history, query JSON, and track token usage through focused tools.

Advanced setup and version management

To skip the wizard, select clients explicitly or configure all supported clients:

npx leantoken setup --claude --codex --yes
npx leantoken setup --all --yes

Use --private-runtime to copy the exact package-native executable into LeanToken's versioned application-data directory and configure clients to launch it directly, without persistent npm/Node wrappers. Preview its path and digest with --dry-run.

Automation never treats detection as consent: --yes requires explicit client flags, --all, or --refresh for entries already managed by LeanToken. Preview the same resolved plan without changing files:

npx leantoken setup --codex --cursor --dry-run

Setup adds the leantoken MCP entry plus a small owned discovery skill in the host-standard user skill directories. The skill advertises routing metadata; it does not duplicate tool schemas, add rules, or install shell hooks. Remove the owned integration with:

npx leantoken remove

Refresh only existing LeanToken MCP entries after explicitly choosing a new version, or use an older version to roll back:

npx --yes leantoken@latest setup --refresh --yes
npx --yes leantoken@0.1.8 setup --refresh --yes --allow-outdated

Common agent workflows

LeanToken works best as a small evidence loop rather than a one-shot repository dump:

1. Orient autonomous triage in one call. Start an uncertain broad task with context and plan_only: false, then use the materialized evidence directly. Make at most one focused follow-up only when coverage identifies a concrete missing implementation or regression-test owner.
2. Continue without resending source. Pass the prior receipt_id on the next context call, or pass returned fragment hashes as known_hashes. The response reports exact and overlapping omissions instead of silently charging the same evidence again.
3. Investigate an observed failure. Use the investigation workflow and provide only directly observed failure_traces, paths, symbols, or test intent in workflow_evidence. Follow with exact search, outline, or read calls for the owners the evidence identifies.
4. Review a change. Use the review workflow with base_revision set to BASE..HEAD and strict_changed_paths: true. Request a handoff when another agent needs a compact manifest of selected hashes, changed paths, assumptions, and completed validations without copied source bodies.

This one-call contract is for autonomous repository triage, not a limit on implementation agents. Human review and control-plane flows can still preview expensive or high-risk retrieval with plan_only: true before materializing. The repeated multi-agent context suite found that an iterative LeanToken profile used 50.9% more total input than thin native, while the frozen one-context-plus-optional-one-search profile saved 20.1% and had 15/20 path-set successes. Those results cover four pinned triage tasks; they do not prove a universal implementation workflow.

Explicit focus constraints are contracts. When a request supplies focus_paths, exact focus_symbols, and minimum_fragments_per_focus_path, LeanToken generates candidates within the documented per-file bounds and reports a coverage failure when distinct ranges cannot satisfy the minimum. Explain-profile plans and materialized responses also identify the bounded allocation boundary that generated, reserved, selected, or suppressed each focus candidate without changing ranking.

Why LeanToken

Most agents start by searching widely and reading whole files. LeanToken narrows that work in stages:

Typical repository exploration	With LeanToken
Scan broad directory listings	Find relevant paths in a compact tree
Read whole files to find structure	See definitions and imports without loading the entire file
Send the same code again after each turn	Avoid repeating unchanged evidence
Let large files fill the request	Keep returned source within an exact source-token budget and report response overhead separately
Guess which files matter	Rank likely relevant code for the task

Your coding agent still handles editing, commands, tests, and conversation. LeanToken finds and returns the code those tasks need.

LeanToken does not create one giant prompt file. It answers focused searches and reads as the agent needs them.

Example

For a task like fix request cancellation during shutdown, an illustrative bounded result might look like this:

Budget: 1,200 source tokens

Selected evidence:
  src/services/executor.rs        lines 137-147, 251-259
  src/services/reconciliation.rs lines 148-175, 257-272

The agent receives these ranges instead of both full files. If the budget is too small, the response also says what was left out. Paths, scores, receipts, JSON, and MCP transport wrappers are not part of this source-token budget; see token accounting for the measured boundaries.

Available tools

Tool	Purpose
leantoken.context	Default materialized first call for autonomous broad triage; optional preview for human or control-plane review.
leantoken.search	Prefer over grep/rg for ranked search; exhaustive text/regex calls can explicitly record or reuse complete query coverage.
leantoken.files	Prefer over find/ls/glob for compact, ignore-aware path discovery.
leantoken.outline	Inspect definitions, signatures, imports, and ranges without whole-file reads.
leantoken.read	Prefer over cat/head/sed for one exact symbol or inclusive line range.
leantoken.history	Read, batch-diff, or trace parsed symbols across immutable Git revisions.
leantoken.json	Query, summarize, or compare bounded live JSON with paged keys and typed diagnostics.
leantoken.receipt_rebase	Explicitly carry only same-path, same-coordinate, same-hash evidence into a newer completed generation.
leantoken.savings	Report observed response accounting, hash suppression, failures, and explicit observation limits.

Advanced retrieval controls

Every index-backed retrieval tool, including receipt_rebase, accepts consistency: "reconcile_working_tree" when completed edits must be reconciled before the query. The default, "indexed_generation", returns the latest completed index generation without scanning or waiting for filesystem changes; it is not a Git revision boundary. leantoken.history reads immutable Git objects and leantoken.json reads exact live files, so neither accepts an index consistency mode. To constrain context to immutable history, pass BASE..HEAD as leantoken.context.base_revision with strict_changed_paths: true.

For autonomous broad triage, set plan_only: false and use the materialized evidence directly. Reserve plan_only: true for human or control-plane review before expensive or high-risk retrieval: it returns bounded ranked candidate metadata without source fragments or receipt mutation. After approval, repeat the same request with plan_only: false. Set response_profile: "compact" for the smallest fail-loud response, keep the default "balanced" shape, or use "explain" for bounded individual omissions, facets, and diff evidence. The response reports the resolved choice as effective_response_profile. Legacy verbose_diagnostics: true maps to "explain" and conflicts with an explicit "compact" or "balanced" profile.

The catalog stays intentionally small because every tool description and schema also consumes model context.

CLI usage

Run LeanToken directly through npx:

npx leantoken status
npx leantoken savings
npx leantoken doctor
npx leantoken --root /path/to/repo search handle_request

Or use a globally installed binary:

npm install --global leantoken@latest

leantoken --root /path/to/repo index
leantoken --root /path/to/repo search handle_request --mode identifier --max-tokens 800
leantoken --root /path/to/repo context \
  --task "fix request cancellation during shutdown" \
  --budget 2000

Audit an existing redacted experiment or host report without opening a repository index:

leantoken episode audit \
  --adapter multi-agent-suite-v1 \
  --input benchmarks/reports/multi-agent-context-suite-v1-codex-0.144.1.json

The default projection is Markdown; add global --json for the stable normalized JSON schema. The auditor is local, bounded, and read-only with respect to its input. It retains artifact hashes, not raw prompts, source, tool arguments, or tool outputs.

npm install leantoken installs the command in the current project's node_modules/.bin; it does not add leantoken to the shell PATH. Invoke a project-local install through npx leantoken, a package script, or ./node_modules/.bin/leantoken.

Run the MCP server manually over stdio:

leantoken --root /path/to/repo mcp

Manual MCP client configuration

{
  "mcpServers": {
    "leantoken": {
      "command": "leantoken",
      "args": ["--root", "/path/to/repo", "mcp"]
    }
  }
}

Installation options

The npm package includes native binaries for:

• macOS on ARM64 and x64
• glibc Linux on ARM64 and x64
• Windows on x64

Installation does not run lifecycle scripts or download an executable from a postinstall hook. Other targets, including musl Linux, must build from source. Install Rust 1.95 or later and a native C/C++ toolchain, then run:

cargo install --git https://github.com/morluto/leantoken

Updating

MCP entries created through npx stay pinned to the exact LeanToken version that configured them. Update existing client integrations explicitly:

npx --yes leantoken@latest setup --refresh --yes

For a globally installed CLI or a CLI installed with Cargo:

leantoken upgrade --check
leantoken upgrade --yes

update is an alias for upgrade. For a project-local npm installation:

npm install leantoken@latest

Pinned MCP entries never silently move to @latest. If the exact package is not available locally or online, startup fails rather than selecting another version. Updating the CLI does not change existing MCP entries. See the usage guide for rollbacks, cache management, and version details.

Cache management

Inspect local repository caches or preview cleanup before applying it:

leantoken cache list
leantoken cache list --summary
leantoken cache list --incompatible-with-current
leantoken cache prune --incompatible-with-current
leantoken cache prune --older-than 30 --dry-run
leantoken cache prune --max-total-bytes 1073741824 --yes

See the usage guide for cache states, pagination, and cleanup safety rules.

How it works

repository
    │
    ▼
file discovery ──► code structure extraction ──► local search index
                                                    │
                                                    ▼
agent request ──► ranked / exact retrieval ──► focused code within a token budget

LeanToken indexes source once, then serves compact paths, ranked matches, structural outlines, exact source ranges, and task-specific context. It avoids resending unchanged evidence across turns.

Dependency-heavy workspaces can opt into a separate, cache-identified first-party index without changing the default whole-repository behavior:

leantoken --index-include 'src/**' --index-include 'tests/**' index

Status and every retrieval disclose whether the active index is full or scoped, so an empty scoped result is never presented as whole-repository absence. See the usage guide for bounds, cache identity, and MCP registration examples.

LeanToken's goal is to return the code an agent needs with fewer input tokens.

Documentation

Guide	Contents
Usage and tool reference	Commands, MCP tools, request options, and examples
Architecture and reliability	Components, data flow, storage, and failure behavior
Roadmap	Current direction and planned work
Development and testing	Local setup, validation, and release workflow
Benchmark methodology	Token-economy measurements and interpretation
Measurement harnesses	Experiment, wire-cost, and profiling tools

License

Licensed under either of the following, at your option:

• Apache License, Version 2.0
• MIT License