Developer Tools●●●Banger
Cobalt – Unit tests for AI agents, like Jest but for LLMs
Jest for LLMs—CI-native eval that fails builds on quality drops, not dashboards.
Ship ItSolve My ProblemBig Brain
fdefitte
303mo ago
Back to browse
GitHub Repository
69 starsTypeScript
We built Cobalt, Open source unit testing for AI Agents
by fdefitte·Feb 12, 2026·3 points·1 comment
AI Analysis
●●SolidSolve My ProblemShip It
Testing framework for AI agents with LLM judges and SQLite result tracking.
Strengths
- •Solves real pain: testing AI agents is under-addressed, and LLM-based evaluation is smart.
- •Good integrations: Langfuse, LangSmith, Braintrust, Basalt tracking built in.
- •MCP server lets AI assistants run experiments, tying into editor workflows naturally.
Weaknesses
- •Crowded space: Braintrust, Humanloop, and LangSmith already offer agent eval.
- •Early stage: 18 stars, limited adoption signals, unclear why this over existing tools.
Category
Target Audience
AI/LLM engineers, agent developers, teams building AI-powered applications
Similar To
Braintrust · LangSmith · Humanloop
Similar Projects
Developer Tools●●●Banger
CheckAgent The open-source pytest testing framework for AI agents
pytest-native testing for AI agents with 101 built-in safety attack probes.
Solve My ProblemSlick
xydac
301mo ago
AI/ML●●Solid
GEDD – A Systematic Evidence Driven LLM as a Judge Framework
Qualitative eval workflow for PMs when LangSmith and Arize target ML engineers.
Big BrainNiche Gem
balasvce2026
203d ago
Developer Tools●●●Banger
Cheddar-bench – unsupervised benchmark for coding agents
Unsupervised bug benchmark using agents as both attackers and defenders—novel scoring methodology.
Big BrainWizardryShip It
przadka
903mo ago
Developer Tools●●Solid
Agent-triage – diagnosis of agent failures from production traces
Replays agent traces step-by-step to pinpoint exact failure turns automatically.
Solve My ProblemBig Brain
oren1531
423mo ago
AI/ML●●●Banger
Meta-agent: self-improving agent harnesses from live traces
Iteratively improves agent harnesses from 67% to 87% on tau-bench using production traces.
Big BrainSolve My Problem
essamsleiman
1402mo ago