{"harness": {"slug": "composio-swekit", "name": "Composio SWEKit", "vendor": "Composiohq", "description": "Composio SWE-kit is Composio's agent scaffold emphasizing tool integrations (third-party APIs, actions) applied to SWE-bench Verified tasks. Public submissions appear under this harness name on swebench.com.\n\nCommunity `swe_bench_site` scores. Useful when studying tool-rich agents vs minimal bash loops \u2014 compare cost and token columns in the table below.", "card_summary": "Composio SWE-kit \u2014 tool-integration agent on Verified.", "homepage_url": "https://composio.dev", "repo_url": "https://github.com/ComposioHQ/composio/tree/master/python/swe/agent", "docs_url": null, "scaffold_code": null, "harbor_agent_name": null, "api_protocol": null, "status": "validated", "supports_rl": true, "score_count": 1, "best_success_rate": 40.6, "github_stars": 29874, "popularity_tier": null, "hrl_score": 33.0, "hrl_rank": 17, "catalog_token_volume": null, "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/logos/20241016_composio_swekit.png", "aa_coding_index": null, "aa_mean_cost_usd": null, "aa_mean_total_tokens": null}, "scores": [{"success_rate": 40.6, "resolved_count": 203, "total_count": 500, "cost_usd": null, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench Verified public submission", "source": "swe_bench_site", "source_url": "https://github.com/ComposioHQ/composio/tree/master/python/swe/agent", "observed_at": "2024-10-16T00:00:00+00:00", "harness": {"slug": "composio-swekit", "name": "Composio SWEKit", "vendor": "Composiohq", "repo_url": "https://github.com/ComposioHQ/composio/tree/master/python/swe/agent", "openrouter_icon_url": null}, "model": {"slug": "claude-3-5-sonnet-20241016", "name": "Claude 3.5 Sonnet (2024-10-16)", "vendor": "Anthropic"}, "benchmark": {"slug": "swe-bench-verified", "name": "SWE-bench Verified"}}], "benchmarks": [{"slug": "swe-bench-verified", "name": "SWE-bench Verified", "category": "eval"}]}