← Back to agents
A

Aider

Agents
85

AI pair programming in your terminal with git integration and multi-file editing

Established terminal-based AI pair programming tool

Metrics

licenseApache-2.0
providerOpen Source
agent typecoding
github stars22,000
base model requirementsclaude-3.5-sonnet, gpt-4, deepseek-coder
tool integrations count12

Score Breakdown

adoption88
tool use88
memory context85
self reflection80
planning reasoning82

Scoring Methodology

planning_reasoning25% weight

Task decomposition and strategic execution capabilities

Source: AgentBench, WebArena, SWE-bench Verified

tool_use25% weight

Intent recognition, function selection, and execution accuracy

Source: ToolBench, API-Bank, Berkeley Function Calling Leaderboard

memory_context20% weight

Long-context handling and information retention

Source: LongBench, RULER, InfiniteBench

self_reflection15% weight

Error recognition and behavior adjustment

Source: ReAct benchmark, Reflexion evaluation

adoption15% weight

GitHub stars, community usage, ecosystem maturity

Source: GitHub API, PyPI downloads, Stack Overflow activity

Last updated: December 24, 2025