Repocode

Repomix is the tool most people will be choosing between. It is more mature than this one, and there are things it does better.

The honest summary: RepoCode's edge is task-scoped ranking and cache-stable output — Repomix doesn't do either today. We don't publish head-to-head retrieval numbers, so weigh that claim (or any tool's accuracy claim, including a competitor's) against your own repository rather than a chart. Repomix is the safer choice on every axis that isn't ranking: it is more mature, better supported, reaches more places, and parses more languages.

If you are already using Repomix and it is working, there is no urgent reason to switch.

1. Code Review Gather just the files modified in a PR, plus their immediate dependencies:

npx repocode . --pr --instruction review --budget 80k

2. Bug Hunting Focus context around specific domain logic:

npx repocode . --query "authentication and session handling" --instruction bugfix --budget 60k

3. Preparing for a new AI Chat Copy a Markdown representation of a feature branch's changes, including relevant dependencies, directly to your clipboard:

npx repocode . --branch feature/auth --format markdown --copy

4. Understand Inclusion Decisions Inspect how candidate files were ranked and which files would be selected:

npx repocode . --explain

Requires Node.js ≥ 20.

# CLI, as a one-off
npx repocode .

# CLI, installed globally
npm install -g repocode

(Note: The following examples assume RepoCode is installed globally; prefix commands with npx for one-off use.)

Token Budgeting & Compression

RepoCode ensures your output fits a specific token limit. When over budget, it compresses files based on their rank, entirely on your machine — the default ceiling (partial) never leaves local:

• tier1: Strips comments and whitespace.
• tier2: Strips function bodies, keeping only structural signatures.
• partial: Trims from the bottom of a file when strictly over budget.

One tier is opt-in and not local: tier3-llm sends selected file contents to your configured LLM provider for a semantic summary (exports/imports/types, with one-line descriptions). It only runs once you've set --llm-provider — with none configured, nothing goes over the network no matter what compression ceiling is in effect. Before using it with sensitive repositories, pair it with an appropriate security mode such as redact, exclude, or fail; see Security and Privacy.

Git and Query Modes

Pack only the files that matter:

repocode . --diff          # staged + unstaged changes
repocode . --branch feat   # branch diff vs auto-detected base
repocode . --pr            # auto-detect PR diff in CI

Re-rank files by relevance to a task using --query:

repocode --query "database migrations" --budget 50k

Interactive & Watch Mode

• Interactive TUI: Run repocode . -i to hand-pick files before packing.
• Watch Mode: Run repocode . -w to keep the context file continuously updated as you type.

Language Support

Import tracing (for intelligent ranking) currently supports: TypeScript, JavaScript, Python, Go, Rust, Ruby, Java, C, C++, PHP, and C#. All other text files are discovered and ranked by baseline heuristics.

RepoCode reads from repocode.config.json in your repository root, allowing you to define defaults, custom instructions, and compression tiers.

Order of precedence: CLI flags > project config > presets > defaults.

No telemetry. RepoCode collects nothing, phones home nowhere, and has no analytics of any kind. Nothing about your code, your usage, or your machine leaves it unless you explicitly ask for a feature that sends data — and there are exactly three, each named below.

• Secret Scanning: RepoCode scans for secrets by default. In MCP mode, it automatically redacts detected secrets before returning output.

• Safe Execution: The CLI and MCP server block arbitrary execution of repository configuration files (e.g., .js configs).

• What can leave your machine, and only if you ask for it:

  • --llm-provider <name> — once set, a file too large to fit locally can be sent to that provider for a tier3-llm summary. The default compression ceiling already permits tier3-llm; what gates the network call is having a provider configured at all, not the ceiling — so --llm-provider is the flag to reach for if you want a guarantee nothing leaves the machine, not --compress.
  • --query-mode semantic — sends file contents to your configured provider for embeddings.
  • --remote <owner/repo> — fetches a repository from GitHub rather than reading a local one.

  Nothing else makes a network request. Everything defaults off, and content returned through MCP or pasted into a hosted assistant is then subject to that client's data-handling policy.

• Vulnerability Reporting: Please report security issues to [email protected].

Usage: repocode [path] [options]

Output
  -f, --format <format>              Output format: xml|markdown|plain|json
  -o, --output <file>                Output file path (default: repocode-output.<ext>)
  --stdout                           Write output to stdout instead of a file
  --copy                             Copy output to clipboard
  --header <text>                    Custom header text prepended to output
  --no-preamble                      Omit AI preamble from output

Budget & compression
  -b, --budget <tokens>              Token budget (e.g. 50k, 100000)
  --model <name>                     Size the budget for a model's context window
                                     (claude-opus-5, gpt-5.6, gemini-2.5-pro, …)
  --no-budget                        Disable token budget — include all files
  --compress <level>                 Compression ceiling: none|tier1|tier2|tier3-llm|partial
  --no-compress                      Disable all compression (alias for --compress none)
  --pin <paths...>                   Always include these files (bypass candidate cut)
  --max-file-size <size>             Skip files larger than this (e.g. 500k, 2m)

File selection
  --include <patterns...>            Include glob patterns
  -e, --exclude <patterns...>        Exclude glob patterns
  --preset <name>                    Activate a named preset from config or built-in
  --list-presets                     List available presets and exit
  --save-preset <name>               Save current CLI flags as a named preset and exit
  --query <text>                     Re-rank files by relevance to this query
  --query-mode <mode>                Query mode: keyword|semantic
  --graph-depth <n>                  Import graph depth for transitive tracing (default: 2)
  --stdin                            Read a newline-separated file list from stdin

Git modes
  --diff                             Uncommitted working-tree changes (staged + unstaged + untracked)
  --staged                           Staged changes only
  --branch <name>                    Diff of <name> vs auto-detected base branch
  --log <n>                          Files touched in the last n commits
  --range <from..to>                 Files changed in a commit range (e.g. HEAD~3..HEAD)
  --pr                               Auto-detect PR base from CI env or remote HEAD
  --base <branch>                    Override base branch for --branch / --pr
  --no-deps                          Disable import tracing for git modes

Remote repositories
  --remote <owner/repo>              Pack a GitHub repo without cloning
  --remote-branch <branch>           Branch to fetch with --remote (default: repo's default branch)
  --remote-token <token>             GitHub token for --remote (overrides GITHUB_TOKEN env var)

Instructions
  --instruction <name>               Use a named instruction from the library
  --instruction-file <path>          Include instruction text from this file
  --instruction-text <text>          Use literal text as the instruction
  --list-instructions                List available instructions and exit

LLM compression
  --llm-provider <name>              openai|anthropic|ollama|custom
  --llm-model <name>                 Model name for the LLM provider
  --llm-budget-tokens <n>            Max tokens per file for LLM compression (default: 8000)

Security
  --no-security                      Disable secret scanning
  --security-mode <mode>             warn|exclude|redact|fail (default: warn)

Other
  --init                             Create a repocode.config.json in the current directory
  --dry-run                          Show what would be packed without writing output
  --explain                          Print ranked file table to stderr and exit (implies --dry-run)
  --verbose                          Show all excluded files in summary
  -i, --interactive                  Open interactive TUI for file selection
  -w, --watch                        Watch for file changes and keep the output file up to date
  --profile <name>                   Load named profile in interactive mode (requires -i)

Commands
  mcp-server                         Start the MCP server (same as `npx -p @repocode/mcp repocode-mcp`)

Exposes RepoCode as Model Context Protocol tools and resources for AI assistants.

{
  "mcpServers": {
    "repocode": {
      "command": "npx",
      "args": ["-y", "-p", "@repocode/mcp", "repocode-mcp"]
    }
  }
}

@repocode/core owns the entire pipeline. repocode (CLI) and @repocode/mcp both import it directly without subprocesses.

Editor integrations are planned.

Contributing

Requires Node.js ≥ 20 and pnpm ≥ 10. See CONTRIBUTING.md.

git clone https://github.com/nextbridgehq/repocode.git
cd repocode
pnpm install
pnpm -r build
pnpm -r test