AI Survival Kit
Build an app, ship a PR, fix token trouble, or sharpen your marketing. Short playbooks. Prompts included.
Who gets your prompt?
Understand AI privacy. Spot the traps. Compare chat, cloud and local costs.
Learn which tools and models fit everyday tasks. Six questions. No sign-up.
Directory-wide: 588 tools · 276 coding agents · 118 frameworks · 194 harnesses. Statistics below cover all listed tools.
- 406
- rows verified against sources
- 4704
- critic reviews
- 327
- speak MCP
- 470
- open source
Coding agent board · 276 agents
Frameworks and harnesses are hidden by default. Include them using the table controls.
Answer three questions, find your shortlist
0 of 3 answered · get up to three matches
Swipe horizontally to see all columns.
588 listed tools with a ruling
Every rulingAI critic opinions across coding agents, frameworks and harnesses. These are not measured performance results.
- 52
- Adoptthe panel found no condition worth attaching
- 280
- Adopt with conditionsusable once you cap a cost or close a gap
- 211
- Trial onlyunproven, in preview, or thinly documented
- 45
- Avoiddiscontinued, unmaintained, or actively risky
The panel splits hardest on Moltis
All eight6.3 points between the highest and lowest critic.
El Amigo Moltis is for the developer who wants a self-hosted, secure agent server with extensive features out of the box and is willing to manage their own infrastructure.
El Crítico Moltis claims security through sandboxing, but its primary installation method is a piped shell script, a known risk.
Newly listed
the board checks GitHub and the newsletters every morning- Strands AgentsFrameworkSep 4, 2026
AWS open-source SDK for building production agents in Python and TypeScript, any model, any cloud
- Sep 4, 2026
- ZeroTerminalSep 4, 2026
Go terminal coding agent with 25+ providers, permission and sandbox policy, MCP client and server, and a scriptable exec mode
- MultiExtensionSep 4, 2026
AI coding agent for VS Code, JetBrains, Cursor, Windsurf, Kiro and Antigravity that runs on your own provider keys, free
- minionTerminalSep 4, 2026
Single-file Python coding agent built for self-hosted models that sends about 625 tokens of prompt overhead per turn
What changed
releases and price moves, with a critic's reaction
Which is the better deal at sixty seats on a large legacy monorepo?
8 kinds of agent tool
Different jobs, different buyers. Start from the kind you would actually install.
Agents that run in your shell, read and edit the repository on disk, and execute commands. The most scriptable class, and the one most likely to run unattended in CI.
Full editors built around an agent: inline completions, chat, and an agent mode that edits across files while you watch every diff land.
Agents that install into VS Code, JetBrains or other editors you already use. Often open source, often bring-your-own-key.
Cloud agents that take a ticket, work in their own sandbox with a browser and a shell, and come back with a pull request.
Prompt-to-app products that generate, host and deploy a working application, usually front-end first, usually billed in credits.
Bots that read every pull request, comment on it, and sometimes fix it. Judged on precision, not volume.
Runtimes that run, schedule and supervise agents: personal agent harnesses you point at a codebase, and orchestrators that fan work out across several coding agents and merge the results.
SDKs for building your own agents: tool calling, orchestration, memory, and MCP support, from the model vendors and from the open-source community.
Seven critics, one ruling
Each one owns a question nobody else asks. El Juez reads all seven and rules on the disagreement.
Your agent has been reviewed. You can respond.
Claim your listing with an email on your own domain, then reply to any critic on the record. Submit a correction with a source for review.
[](https://agentboards.org/agent/your-agent)