ai-gent Manifest
This document defines the principles, boundaries, and governance philosophy of ai-gent.
What ai-gent does — and what it deliberately avoids
ai-gent is built for people and organizations that need certainty before trust.
It deliberately avoids
- No training on your private conversations
- No hidden data flows or silent telemetry
- No mandatory cloud or closed APIs
- No black-box governance or retroactive rule changes
- No growth-at-all-costs incentives
Instead
- Agents run where you decide
- Preference for the smallest viable models to reduce data exposure and resource use
- Clear, published rules for agents, monetization, and governance
ai-gent prioritizes long-term trust, privacy, and sustainability over short-term adoption.
How ai-gent agents run
Local or controlled execution
Agents can run locally, on your own infrastructure, or in explicitly chosen environments — never silently elsewhere.
Smallest viable models first
Agents are designed to work with minimal, task-appropriate models to reduce data exposure, costs, and energy usage.
Composable and replaceable
Each agent is modular. Models, tools, and data sources can be swapped without rewriting the system.
Why ai-gent uses a curated agent store
Open ecosystems fail when trust and incentives are undefined. ai-gent.store exists to make rules explicit — before scale.
- No unreviewed or opaque agents
- Every published agent follows transparent rules for data handling, execution, and dependencies
- Clear monetization without dark patterns
- Pricing, revenue sharing, and obligations are defined upfront — no hidden fees, no retroactive changes
- Accountability over virality
- Agents are curated for clarity, maintainability, and responsible behavior, not engagement metrics
- Governance before growth
- Publication and removal rules are documented and enforced openly
The store is not an app marketplace. It is a trust boundary.
ai-gent is built in the open because trust cannot be retrofitted.
- Rules before power: governance rules are defined and published before influence or scale
- Decisions are documented: changes to policies, store rules, and direction are discussed and recorded publicly
- No single controlling entity: ai-gent is structured to avoid unilateral control over agents, data, or monetization
- Participation is optional, transparency is not: you don’t need to engage in governance — but you can always inspect it
The community exists to constrain power, not to create consensus theater.