Choose an agent for the work—and see the authority it needs
Collections organize candidate repositories around a real operator decision. Security scores still come only from each repository’s reviewed canonical result; popularity and collection membership never add points.
Comparison gateA “best” or ranked use-case conclusion requires at least two eligible, reviewed results under a comparable baseline.
Cohort building0/2 reviewed minimum
Always-On Business Operator
Long-running personal and business agents with memory, scheduling, messaging, remote control, and broad integration access.
Decision
Which open-source personal agent is the most defensible foundation for an always-on business operator?
Security question
Which agent has the strongest gateway, memory, credential, scheduling, and remote-control boundaries?
Cohort building0/2 reviewed minimum
Build and Ship Software
Coding agents and engineering platforms that can inspect repositories, change code, run commands, and participate in delivery workflows.
Decision
Which open-source coding agent is the best security fit for a small engineering team or solo founder?
Security question
Which coding agent can safely receive repository, shell, CI, cloud, and deployment access?
Cohort building0/2 reviewed minimum
Build Your Own Agent Platform
Frameworks and extensible platforms used to assemble custom agents, tools, sandboxes, and orchestration.
Decision
Which open-source agent foundation gives builders the clearest security and evidence controls?
Security question
Which platform exposes usable isolation, permission, extension-trust, and evidence controls?
Cohort building0/2 reviewed minimum
Marketing and Sales
Agents and automations with prospect data, social accounts, email, content, and outbound authority.
Decision
Which open-source marketing or sales agent can be safely limited to drafts and approved outbound actions?
Security question
Which tools can be trusted with social accounts, prospect data, Gmail, and outbound actions?
Cohort building0/2 reviewed minimum
One-Person Business Stack
A complete solo-business workflow spanning research, product planning, code, deployment, marketing, and human approval.
Decision
Can a small AI-agent stack help one person launch a business without giving every component every credential?
Security question
Can the workflow prevent one compromised component from inheriting every credential and consequential action path?
Cohort building0/2 reviewed minimum
Research and Browser Automation
Agents that retrieve untrusted content and operate browsers, research tools, or authenticated web sessions.
Decision
Which research or browser agent best contains hostile content while completing useful authenticated work?
Security question
Can untrusted webpages or retrieved content influence an authenticated browser or downstream agent?
Composition evidence is separate
Repository posture does not prove a stack is safe.
Exact, pinned combinations are assessed through scenario-based workflow testing after the isolated laboratory passes its validation gates.