# Working state

The current, inspectable record of an agent’s task position, observations, decisions, pending actions, and budgets.

technical-term · Context & knowledge · maturing · Reviewed 2026-08-09

## Definition

Working state is the operational state needed to continue an agent run safely: goal, plan, completed steps, tool results, unresolved questions, checkpoints, resource budgets, verifier dispositions, and terminal status. It should be explicit and durable where recovery matters.

State-machine techniques are established; consistent agent-state models and portability are still emerging.

## Why it matters

State should survive model calls and, for long-running work, process or context loss.

A transcript is useful evidence but often a poor canonical state representation.

## System anatomy

- **Task position:** What has completed and what remains.
- **Observations:** Validated results from tools and the environment.
- **Budgets:** Remaining time, steps, spend, and action limits.
- **Disposition:** Running, blocked, held, escalated, completed, or failed.



## Important distinctions

- **Conversation history:** History records exchanges; working state is a structured, current operating record.
- **Durable memory:** Working state belongs to a run; durable memory can outlive it and requires separate governance.



## Implementation signals

- Use typed, versioned state transitions
- Checkpoint before irreversible or long-running actions
- Separate observed facts from model hypotheses

## Failure modes

- The model reconstructs critical state from a truncated transcript
- Two agents update shared state without concurrency control



## Related knowledge

- [Agent loop](https://aisdlc.ai/agentic-engineering/agent-loop) — The finite runtime cycle in which an agent interprets state, selects an action, observes the result, updates state, and decides what happens next.
- [Durable project memory](https://aisdlc.ai/agentic-engineering/durable-memory) — Persistent, attributable project knowledge that carries decisions, outcomes, requirements, failures, and operating state across agent sessions without assuming that every stored item remains true or safe.
- [Durable execution](https://aisdlc.ai/agentic-engineering/durable-execution) — Execution that persists state and can resume safely across process, model, context, worker, or environment interruptions.
- [Evidence engineering](https://aisdlc.ai/agentic-engineering/evidence-engineering) — The design of versioned, queryable evidence linking requirements, decisions, implementations, tests, evaluations, approvals, deployments, runtime signals, and lifecycle actions.

## Sources and further study

- [Anthropic — Effective harnesses for long-running agents](https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents) — Patterns for incremental progress and durable handoffs across context windows and extended execution. **Use:** First-party case study. **Limitation:** This first-party account documents one organization, product, or implementation context and should not be generalized without local evidence.
- [Anthropic — Scaling Managed Agents: Decoupling the brain from the hands](https://www.anthropic.com/engineering/managed-agents) — An architecture separating durable sessions and environment state from changing models and harnesses. **Use:** First-party case study. **Limitation:** This first-party account documents one organization, product, or implementation context and should not be generalized without local evidence.
- [OpenAI — An open-source spec for Codex orchestration: Symphony](https://openai.com/index/open-source-codex-orchestration-symphony/) — An orchestration specification that connects project work, isolated agent execution, and reviewable delivery state. **Use:** First-party case study. **Limitation:** This first-party account documents one organization, product, or implementation context and should not be generalized without local evidence.

---

This library synthesizes cited research, standards, official documentation, and clearly attributed practitioner perspectives. Maturity describes the state of a concept—not vendor endorsement, production readiness, or permission to deploy. Benchmarks and demonstrations do not replace use-case evaluation, governed controls, independent verification, or named human release authority.