Tool to benchmark AGENTS.md file on swe tasks

emiliolugo1 pts0 comments

GitHub - emiliolugo/clawmark: clawmark runs standardized benchmarks on agent configs · GitHub

/" data-turbo-transient="true" />

Skip to content

Search or jump to...

Search code, repositories, users, issues, pull requests...

-->

Search

Clear

Search syntax tips

Provide feedback

--><br>We read every piece of feedback, and take your input very seriously.

Include my email address so I can be contacted

Cancel

Submit feedback

Saved searches

Use saved searches to filter your results more quickly

-->

Name

Query

To see all available qualifiers, see our documentation.

Cancel

Create saved search

Sign in

/;ref_cta:Sign up;ref_loc:header logged out"}"<br>Sign up

Appearance settings

Resetting focus

You signed in with another tab or window. Reload to refresh your session.<br>You signed out in another tab or window. Reload to refresh your session.<br>You switched accounts on another tab or window. Reload to refresh your session.

Dismiss alert

{{ message }}

emiliolugo

clawmark

Public

Notifications<br>You must be signed in to change notification settings

Fork

Star

main

BranchesTags

Go to file

CodeOpen more actions menu

Folders and files<br>NameNameLast commit message<br>Last commit date<br>Latest commit

History<br>38 Commits<br>38 Commits

.claude

.claude

.github/workflows

.github/workflows

data

data

docs

docs

src

src

tests

tests

.gitignore

.gitignore

AGENTS.md

AGENTS.md

CLAUDE.md

CLAUDE.md

CONTRIBUTING.md

CONTRIBUTING.md

Cargo.lock

Cargo.lock

Cargo.toml

Cargo.toml

LICENSE

LICENSE

PLAN.md

PLAN.md

README.md

README.md

View all files

Repository files navigation

clawmark

Local Rust CLI to compare agent instruction files (CLAUDE.md / AGENTS.md variants) on real code.

clawmark runs two or more variant files against a bundled 5-task SWE-bench Lite smoke set, invokes Claude or Cursor per variant, evaluates patches with the official SWE-bench harness in Docker, and prints a leaderboard.

Commands: doctor · run · report

Quickstart

1. Check prerequisites

cargo run -- doctor

You need Rust, Docker, Python 3.11+, swebench, and git. For agents: claude and/or cursor-agent on PATH (only the backend you use must be installed and authenticated).

2. Create variant files (inside your working directory)

variants/a.md<br>echo "Make the smallest fix. Prefer targeted tests only." > variants/b.md">mkdir -p variants<br>echo "Make the smallest fix. Do not run the full test suite." > variants/a.md<br>echo "Make the smallest fix. Prefer targeted tests only." > variants/b.md

3. Run the benchmark

cargo run -- run \<br>--a variants/a.md \<br>--b variants/b.md \<br>--model sonnet \<br>--timeout-secs 600 \<br>--out out

This runs 2 variants × 5 tasks = 10 agent invocations . --out must not already exist.

4. Read the report (also printed at the end of run)

cargo run -- report --out out<br>cargo run -- report --out out --show-patches # include patches (20 lines each)

Common variations

Different model per variant:

cargo run -- run --a variants/a.md --b variants/b.md \<br>--model sonnet --model-b haiku --timeout-secs 600 --out out

Claude vs Cursor (A=claude, B=cursor):

cargo run -- run --a variants/a.md --b variants/b.md \<br>--model sonnet --agent-b cursor --timeout-secs 600 --out out

Three or more variants:

cargo run -- run \<br>--variant alpha=variants/a.md \<br>--variant beta=variants/b.md \<br>--variant gamma=variants/c.md \<br>--variant-model alpha=sonnet \<br>--variant-model beta=haiku \<br>--variant-model gamma=sonnet \<br>--timeout-secs 600 \<br>--out out

Concurrent tasks within a variant pass:

cargo run -- run --a variants/a.md --b variants/b.md \<br>--model sonnet --timeout-secs 600 --parallel 3 --out out

Release binary:

cargo build --release<br>./target/release/clawmark run --a variants/a.md --b variants/b.md --model sonnet --timeout-secs 600 --out out

Documentation

What it does

CLI reference

Output layout

Notes

Contributing

See CONTRIBUTING.md. Open an issue before large changes.

License

MIT

About

clawmark runs standardized benchmarks on agent configs

Resources

Readme

License

MIT license

Contributing

Contributing

Uh oh!

There was an error while loading. Please reload this page.

Activity

Stars

stars

Watchers

watching

Forks

forks

Report repository

Releases

No releases published

Packages

Uh oh!

There was an error while loading. Please reload this page.

Contributors

Uh oh!

There was an error while loading. Please reload this page.

Languages

Rust<br>100.0%

You can’t perform that action at this time.

variants cargo variant model claude clawmark

Related Articles