RepoHarvester
English | Русский
Clone a git repository (or point at a local folder) and get all of its text files in one file. Paste it into ChatGPT or Claude as LLM context, search it, review it, or read it offline. A command-line tool plus an optional GUI. No runtime dependencies.
uvx repoharvester https://github.com/psf/requests
Cloning https://github.com/psf/requests
Wrote 92 files to output/requests_all_files.txt, ~154.7k tokens (0 non-UTF-8 skipped)
Features
- Works with SSH/HTTPS git URLs (shallow clone into a temp dir) or local folders.
- Skips binary/noise by default: media, archives, executables, lock files,
node_modules, virtualenvs, caches,.git,.github, and more. - Optional comment removal for common languages. String literals are kept, so
"http://..."and"#fff"stay intact. Python uses the real tokenizer; other languages use regexes. - Size limit per file. Non-UTF-8 files are skipped and listed in
skipped_files.txt. - Respects
.gitignore: in a git repo only tracked and untracked-but-not-ignored files are harvested. - Logs a rough token estimate (~4 characters per token) so you know if it fits an LLM context.
- Each file is headed by its path relative to the repo root. The output folder is never harvested into itself.
- No runtime dependencies. The GUI is optional.
Install
Requires Python 3.12+ and git.
uv tool install repoharvester # or: pipx install repoharvester
uv tool install "repoharvester[gui]" # with the GUI
Usage
repoharvester <repo-url-or-folder> [options]
| Option | Description |
|---|---|
-r, --remove |
Remove comments from code files |
--no-skip GROUP[,GROUP] |
Include groups that are skipped by default (repeatable) |
--max-size KB |
Skip files larger than this (default 1000). Files over 500 KB are logged |
--exclude DIR[,DIR] |
Folder names to exclude, along with their contents (repeatable) |
--include GLOB[,GLOB] |
Only harvest files matching these globs, e.g. *.py,*.md (repeatable) |
--exclude-glob GLOB[,GLOB] |
Skip files matching these globs, e.g. tests/* (repeatable) |
-o, --output-dir |
Output directory (default output), or - to print to stdout |
--log FILE |
Also write the log to a file |
Groups: archive, audio, cache_files, certificate, compiled_code, configuration, database, executables, font, log_files, media, node_modules, office, package_locks, system, temporary, video, virtual_env. See EXTENSION_GROUPS, FILENAME_GROUPS and FOLDER_GROUPS in repoharvester.py.
Example:
repoharvester git@github.com:username/repo.git -r --no-skip configuration --max-size 250 --exclude vendor,docs
# only Python and Markdown, no tests, straight to the clipboard
repoharvester . --include '*.py,*.md' --exclude-glob 'tests/*' -o - | xclip -selection clipboard # macOS: | pbcopy
Globs match the path relative to the repo root or the bare file name; * also crosses /, so write test_*.py or docs/* (not **/test_*.py or docs/). Output to stdout is always UTF-8. They apply on top of the groups and .gitignore.
GUI
repoharvester-gui # after installing with [gui]
uv run --extra gui repoharvester-gui # from a clone
Paste a git URL or choose a folder, tick anything extra to include, press Harvest (or Enter), then Open file.
While a clone runs the button turns into Cancel. Settings are remembered in ~/.config/repoharvester/gui.json.
On Linux the project prefers the system Python: uv's bundled builds ship Tk without font antialiasing.
Output format
## <repo-name>
### Files: <count>, ~<tokens> tokens
<relative/path/one.ext> ~<tokens>
<relative/path/two.ext> ~<tokens>
### <relative/path/one.ext>
<file contents>
### end of file
...
The file list up front shows the model the repository's shape and shows you which files eat the context. Token counts are estimates (~4 characters per token).
Written to <output-dir>/<repo-name>_all_files.txt, with skipped_files.txt next to it.
Alternatives
- repomix (Node.js): more output formats, exact token counts, many more options. Pick it if you have Node and want those.
- gitingest (Python): also has a website where you swap
hubforingestin a GitHub URL. - files-to-prompt (Python): local files and folders, no cloning.
RepoHarvester is a single standard-library module: nothing to install besides Python and git, plus a small desktop GUI if you prefer clicking.
Development
See CONTRIBUTING.md.
uv run --group dev pytest
uvx ruff check .
License
MIT. See LICENSE.
Metadata
Release files for repoharvester 0.3.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| repoharvester-0.3.1.tar.gz | 178.7 kB | With provenance | Details for repoharvester-0.3.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| repoharvester-0.3.1-py3-none-any.whl | Python 3 | none | any | Details for repoharvester-0.3.1-py3-none-any.whl |
Total release size: 194.0 kB
Release files / repoharvester-0.3.1.tar.gz
| Download URL | repoharvester-0.3.1.tar.gz |
|---|---|
| Size | 178.7 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
f09205031050588fedae4dc3ad820d1fdf369ff35df312f1937f75fe9cc41bc5
|
|
BLAKE2b-256 checksum How to use checksums |
966deee607a7a4e5e0473577368ae5c1e115c8e8f8a29bae1f2ad20b22eff4e3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 15, 2026.
Transparency logRelease files / repoharvester-0.3.1-py3-none-any.whl
| Download URL | repoharvester-0.3.1-py3-none-any.whl |
|---|---|
| Size | 15.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
95cf15b1f0cf5050855bf3e4b5d304b0dd062446c718abdeff76a6cb21f55d5f
|
|
BLAKE2b-256 checksum How to use checksums |
80b6223fb0cf58644a7f2f3fc34bc88ccf868dafad9ff5219a1e9bdbf5c836c8
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 15, 2026.
Transparency logSecurity for repoharvester 0.3.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for repoharvester-0.3.1.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 15, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.