Report · ai-coding-agents · updated 2026-09-24
AI coding agents
Launch map of products that plan, edit, and open pull requests with an agent loop. Entities are included only when a public, dated source exists. This is not a ranking of model quality; it is an evidence pack for Jev to score on proven execution and validated direction.
Placements
10 of 10 entities have a Jev placement. Axes: Proven Execution × Validated Direction.
| Entity | Status | Proven Execution | Validated Direction | uX | uY | Region | P(region) |
|---|---|---|---|---|---|---|---|
| Cursor | scored | 0.757 | 0.578 | 0.265 | 0.293 | OPERATIONAL | ANCHORED:0.38 OPERATIONAL:0.58 DIRECTED:0.03 FORMING:0.01 |
| GitHub Copilot | scored | 0.425 | 0.465 | 0.356 | 0.321 | OPERATIONAL | ANCHORED:0.18 DIRECTED:0.15 OPERATIONAL:0.60 FORMING:0.07 |
| Claude Code | scored | 0.583 | 0.432 | 0.340 | 0.368 | OPERATIONAL | DIRECTED:0.07 ANCHORED:0.19 FORMING:0.02 OPERATIONAL:0.72 |
| OpenAI Codex | scored | 0.277 | 0.313 | 0.327 | 0.342 | OPERATIONAL | FORMING:0.23 OPERATIONAL:0.51 ANCHORED:0.05 DIRECTED:0.21 |
| Devin | scored | 0.500 | 0.478 | 0.308 | 0.323 | OPERATIONAL | DIRECTED:0.32 ANCHORED:0.09 FORMING:0.09 OPERATIONAL:0.50 |
| Cline | scored | 0.223 | 0.363 | 0.334 | 0.370 | OPERATIONAL | OPERATIONAL:0.50 ANCHORED:0.01 FORMING:0.34 DIRECTED:0.15 |
| OpenCode | scored | 0.585 | 0.350 | 0.322 | 0.334 | OPERATIONAL | ANCHORED:0.16 FORMING:0.02 OPERATIONAL:0.74 DIRECTED:0.08 |
| Amp | scored | 0.295 | 0.457 | 0.300 | 0.270 | DIRECTED | OPERATIONAL:0.30 DIRECTED:0.35 FORMING:0.33 ANCHORED:0.02 |
| Google Antigravity | scored | 0.323 | 0.532 | 0.333 | 0.293 | OPERATIONAL | OPERATIONAL:0.47 DIRECTED:0.32 ANCHORED:0.16 FORMING:0.05 |
| Kiro | scored | 0.400 | 0.482 | 0.350 | 0.306 | DIRECTED | OPERATIONAL:0.08 ANCHORED:0.14 FORMING:0.04 DIRECTED:0.74 |
Entities
- Cursor
Agent-first editor and cloud agent platform from Anysphere.
4 evidence · scored · OPERATIONAL
- GitHub Copilot
GitHub’s coding agent and Agent HQ surface inside GitHub and VS Code.
3 evidence · scored · OPERATIONAL
- Claude Code
Anthropic’s agentic coding tool for terminal, IDE, and GitHub.
2 evidence · scored · OPERATIONAL
- OpenAI Codex
OpenAI’s coding agent, available through GitHub Agent HQ and ACP-compatible editors.
2 evidence · scored · OPERATIONAL
- Devin
Cognition’s cloud and desktop agent platform; Windsurf is now Devin Desktop.
2 evidence · scored · OPERATIONAL
- Cline
Open-source, model-agnostic coding agent with a desktop app for open-weight models.
1 evidence · scored · OPERATIONAL
- OpenCode
Open-source coding agent for terminal, IDE, and desktop.
2 evidence · scored · OPERATIONAL
- Amp
Frontier coding agent and remote orb environment, spun out of Sourcegraph.
3 evidence · scored · DIRECTED
- Google Antigravity
Google’s agent-first development platform and Antigravity CLI, successor to Gemini CLI for individuals.
2 evidence · scored · OPERATIONAL
- Kiro
AWS’s spec-driven agentic IDE and CLI; successor path for Amazon Q Developer IDE plugins.
2 evidence · scored · DIRECTED
Payment does not affect scores. Submit evidence or read the rubric.