# aider-ai-aider > A terminal pair-programming assistant: you add files from your git repo to the chat, describe a change or bug fix, and the LLM edits the code with SEARCH/REPLACE diffs, auto-commits the change, and can re-run your linter/tests to fix failures. - repo: https://github.com/Aider-AI/aider - commit: 5dc9490bb35f9729ef2c95d00a19ccd30c26339c - api style: openai - runs: 19 (None with a reward) - tasks tried: 5 - models: bedrock/us.anthropic.claude-haiku-4-5-20251001-v1:0 - domains: swe - languages: cpp, css, go, html, java, javascript, php, python, ruby, rust - capabilities: edits-files, runs-shell, runs-tests, uses-git, browses-web, reads-docs ## How it runs here Aider is a terminal AI pair-programming harness written in Python (package `aider-chat`, console entrypoint `aider = aider.main:main`). It edits files in the current working directory (using git for its repo map and change tracking) and makes every LLM call through litellm: `aider/llm.py` lazy-loads litellm and `Model.send_completion` in `aider/models.py` calls `litellm.completion` with the model name, streaming flag and optional tool spec. Using an `openai/` model makes litellm's OpenAI chat-completions client honour `OPENAI_API_BASE`/`OPENAI_BASE_URL`, so all requests (main, weak and editor models) go to `$PROXY_URL/v1/chat/completions`. A single non-interactive task is run with `--message "$TASK"` plus `--yes-always`, which answers every confirmation, applies the edits and exits after one turn; chat and full LLM request/response history files are written under /out. ## Results by task | task | runs | last reward | best reward | last tests | |---|---|---|---|---| | [aider_polyglot/polyglot_go_palindrome-products](https://harnessreport.com/tasks/aider_polyglot/polyglot_go_palindrome-products.md) | 2 | 0 | | | | [aider_polyglot/polyglot_python_bowling](https://harnessreport.com/tasks/aider_polyglot/polyglot_python_bowling.md) | 12 | 1 | | 31 passed | | [bird-bench/card_games__372](https://harnessreport.com/tasks/bird-bench/card_games__372.md) | 1 | 1 | | | | [humanevalfix/python-12](https://harnessreport.com/tasks/humanevalfix/python-12.md) | 1 | 1 | | 1 passed | | [swebench-verified/psf__requests-5414](https://harnessreport.com/tasks/swebench-verified/psf__requests-5414.md) | 3 | 0 | | 1 failed, 130 passed, 1 xfailed, 158 errors | ## Tests to run next _ranked by llm_ | task | why | |---|---| | [aider_polyglot/polyglot_cpp_allergies](https://harnessreport.com/tasks/aider_polyglot/polyglot_cpp_allergies.md) | Establish C++ support as a covered compiled language in your core polyglot test suite. | | [aider_polyglot/polyglot_javascript_triangle](https://harnessreport.com/tasks/aider_polyglot/polyglot_javascript_triangle.md) | Extend to JavaScript by solving a similar geometric problem as your passing bowling task to verify dynamic-language support. | | [aider_polyglot/polyglot_rust_bowling](https://harnessreport.com/tasks/aider_polyglot/polyglot_rust_bowling.md) | Confirm Rust support by tackling the same bowling-scoring logic in a systems language, testing cross-language consistency. | | [quixbugs/quixbugs-python-sieve](https://harnessreport.com/tasks/quixbugs/quixbugs-python-sieve.md) | Test single-line bug-fixing in a new taskset to validate that focused editing is a core pair-programming capability. | | [humanevalfix/python-11](https://harnessreport.com/tasks/humanevalfix/python-11.md) | Assess function-fixing consistency on python-11 now that python-12 passed, confirming this pattern is replicable beyond single trials. | ## Runs | run | task | verifier says | calls | seconds | |---|---|---|---|---| | [20260925T090301-aider-ai-aid-psf__requests-5414](https://harnessreport.com/runs/20260925T090301-aider-ai-aid-psf__requests-5414.md) | swebench-verified/psf__requests-5414 | reward 0 · 1 failed, 130 passed, 1 xfailed, 158 errors · verifier exited 1 | 3 | 23 | | [20260925T090255-aider-ai-aid-polyglot_go_palindrome-products](https://harnessreport.com/runs/20260925T090255-aider-ai-aid-polyglot_go_palindrome-products.md) | aider_polyglot/polyglot_go_palindrome-products | reward 0 | 1 | 12 | | [20260924T225830-sw-aider-psf__requests-5414](https://harnessreport.com/runs/20260924T225830-sw-aider-psf__requests-5414.md) | swebench-verified/psf__requests-5414 | reward 0 · 8 failed, 123 passed, 1 xfailed, 158 errors · verifier exited 1 | 2 | 23 | | [20260924T225115-sw-aider-card_games__372](https://harnessreport.com/runs/20260924T225115-sw-aider-card_games__372.md) | bird-bench/card_games__372 | reward 1 | 1 | 21 | | [20260924T224953-sw-aider-python-12](https://harnessreport.com/runs/20260924T224953-sw-aider-python-12.md) | humanevalfix/python-12 | reward 1 · 1 passed | 1 | 16 | | [20260924T224501-sw-aider-psf__requests-5414](https://harnessreport.com/runs/20260924T224501-sw-aider-psf__requests-5414.md) | swebench-verified/psf__requests-5414 | reward 0 · 8 failed, 123 passed, 1 xfailed, 158 errors · verifier exited 1 | 3 | 30 | | [20260924T224300-sw-aider-polyglot_go_palindrome-products](https://harnessreport.com/runs/20260924T224300-sw-aider-polyglot_go_palindrome-products.md) | aider_polyglot/polyglot_go_palindrome-products | reward 0 | 1 | 12 | | [20260924T184611-aider-ai-aid-polyglot_python_bowling](https://harnessreport.com/runs/20260924T184611-aider-ai-aid-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 1 · 31 passed | 1 | 19 | | [20260924T183432-polyglot_python_bowling](https://harnessreport.com/runs/20260924T183432-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 1 · 31 passed | 1 | 21 | | [20260924T180336-aider-ai-aid-polyglot_python_bowling](https://harnessreport.com/runs/20260924T180336-aider-ai-aid-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 1 · 31 passed | 1 | 19 | | [refactorcheck-polyglot_python_bowling](https://harnessreport.com/runs/refactorcheck-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 1 · 31 passed | 1 | 21 | | [20260924T170312-aider-ai-aid-polyglot_python_bowling](https://harnessreport.com/runs/20260924T170312-aider-ai-aid-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 1 · 31 passed | 1 | 20 | | [20260924T162919-aider-ai-aid-polyglot_python_bowling](https://harnessreport.com/runs/20260924T162919-aider-ai-aid-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 0 · 25 failed, 6 passed | 1 | 18 | | [20260924T162753-aider-ai-aid-polyglot_python_bowling](https://harnessreport.com/runs/20260924T162753-aider-ai-aid-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 0 · 25 failed, 6 passed | 1 | 19 | | [20260924T162614-aider-ai-aid-polyglot_python_bowling](https://harnessreport.com/runs/20260924T162614-aider-ai-aid-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 0 · 25 failed, 6 passed | 1 | 18 | | [20260924T162204-aider-ai-aid-polyglot_python_bowling](https://harnessreport.com/runs/20260924T162204-aider-ai-aid-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 0 · 25 failed, 6 passed | 1 | 20 | | [20260924T161445-amd64-aider-polyglot_python_bowling](https://harnessreport.com/runs/20260924T161445-amd64-aider-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 0 · 26 failed, 5 passed | 1 | 18 | | [20260924T160732-amd64-aider-polyglot_python_bowling](https://harnessreport.com/runs/20260924T160732-amd64-aider-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 0 · 25 failed, 6 passed | 1 | 19 | | [20260923T112810-aider-ai-aider-polyglot_python_bowling](https://harnessreport.com/runs/20260923T112810-aider-ai-aider-polyglot_python_bowling.md) | aider_polyglot/polyglot_python_bowling | reward 1 · 31 passed | 1 | 17 | --- Harness Report runs agent harnesses from their GitHub repos on Harbor tasks and records every model call. Every page is also `.md` and `.json`; index: https://harnessreport.com/llms.txt · MCP: https://harnessreport.com/mcp