Agent Atlas / Agent frameworks

mastra-ai/mastra

A TypeScript framework for agents and AI applications with model routing, graph workflows, suspend-and-resume control, memory, MCP, evaluation, and observability.

Agent framework Scope profile Project documented Reviewed August 26, 2026
Intended operator

TypeScript developers building and deploying agent applications.

Primary surfaces

TypeScript framework / workflow engine / studio and deployment adapters

Tracked repository

mastra-ai/mastra

Teaching mechanism

Suspension persists workflow state before human input

The project documents human-in-the-loop suspension backed by storage so a workflow can pause and later resume from durable execution state.

Evidence limitTheGitReporter has not yet inspected storage adapters, failure recovery, or enforcement across deployments.
Inspect the source
System map status

Reporting is still in progress.

This scope profile names the project's place in the field and one useful mechanism. It does not claim that every runtime boundary has been inspected.

Learn Through This Project

8 lessons
Chapter 1 Before the First Model Request Chapter 3 Tool Calls: Contract, Authorization, Execution Chapter 4 Loops, Workflows, and Recovery Chapter 5 Delegation Is a Distributed System Chapter 6 Memory Is a Typed Storage Contract Chapter 7 The Approval Trap Chapter 9 Trust Must Make Its Case Chapter 10 The Future Has Three Branches

Compare Its Boundaries

7 labs
Comparison labTool ContractsComparison labGoverned MemoryComparison labDelegation ReceiptsComparison labApproval BoundariesComparison labSessions And RecoveryComparison labEvidence And EvaluationComparison labMaturity Signals

Latest Reporting

7 articles
Builder Work: Mastra Tests Memory Prompts Against Their Own Noise

September 1, 2026 — Mastra’s new memory simulator freezes the data-generating boundary, isolates each prompt arm, and makes a same-prompt A/A run expose the noise that a flashy A/B result might otherwise hide. mastra-ai/mastra

 A Sandbox Isn’t Finished Until the Repository Says So

August 16, 2026 — Mastra Factory is making one understated agent-system choice explicit: before a session workspace is scrubbed, pooled, or destroyed, the repository can say how its own worktree should be put away. mastra-ai/mastra

 In Mastra’s Approval Queue, the Second Yes Needs Its Own Address

August 10, 2026 — An open Mastra core pull request treats a human approval as a per-tool-call decision, not a vague signal attached to an entire agent run. mastra-ai/mastra

 Mastra gives recurring agent reminders a way to disappear

August 6, 2026 — Mastra’s new transient-signal path lets a processor keep a live instruction near the model without quietly turning that instruction into remembered conversation—and its public commit shows coding agents in the work. mastra-ai/mastra

 API Discovery: Make the Model Own Its Web Search

July 30, 2026 — Mastra’s new webSearchTool is deliberately not a generic search client: it selects a provider-native tool from the active model or fails where the provider boundary is unclear. mastra-ai/mastra

 Builder Work: A Sandbox Is Reusable Only After It Forgets

July 30, 2026 — Mastra’s public sandbox-pool patch treats reuse as a recovery problem: clean the worktree, separate credentials, and assume the machine may fail before it can be claimed again. mastra-ai/mastra

 Mastra: turning TypeScript application primitives into agent infrastructure

July 26, 2026 — Mastra's open source says the difficult part of an agent stack is not producing a turn—it is making state, resumption, tracing, and tool authority behave like application concerns. mastra-ai/mastra

Primary Sources

Mastra repository

Primary public framework and documentation source.

Open Questions

  • Which pause, approval, and recovery guarantees survive across storage and deployment adapters?

Evidence state describes what this Atlas profile can support today. It is not a product score and it is not evidence that uninspected capabilities are absent.