promise-theory

Promise Theory

Promise theory (Mark Burgess; formalized with Jan Bergstra) is a method of analysis for systems of autonomous agents — humans, LLM agents, APIs, and deterministic automation. It supplies the vocabulary for designing and diagnosing delegation: promises, acceptances, assessments, breaches, and renegotiation. This skill is a thin router; load the dense material only when a row in Load By Need matches your task.

Core model

A promise is an autonomous declaration of intended, but as yet unverified, behaviour from a promiser to a promisee (body: label Λ, type τ, constraint χ). Agents are autonomous: no agent can promise another's behaviour. Coordination emerges from voluntary cooperation — an offer plus an acceptance (a counter-promise) — never from imposed obligation. Obligations are derived, non-autonomous impositions (imposition + penalty). Agents keep promises via an evaluation loop: observe → assess → act, converging on the promised state. The Downstream Principle: the most downstream party in a promise chain carries the greatest causal responsibility for the outcome.

When to use

Load this skill when any of these triggers matches:

• Modeling delegation between humans and agents — decide who may promise what to whom, and who accepts, in a human + AI workforce.
• Designing capability manifests or agent contracts — declare capabilities and intent with acceptance criteria, verification, and withdrawal semantics.
• Diagnosing coordination failures — explain unkept promises, refused acceptances, or missing assessments in multi-agent work.
• Calibrating trust and verification — decide how much to verify an agent, at what rate, and at what cost.
• Designing self-healing or convergent infrastructure — evaluation loops that observe, assess, and act toward a desired state.
• Converting obligation-based designs to promise-based ones — replace push commands and mandates with voluntary offers and acceptance.

When not to use