HarnessMatch
Menu

Coding agent

Cursor CLI

Cursor's familiar agent loop in the terminal and in automation.

Status
Active
License
Proprietary
Evidence
Documented + independently measured configuration, 15 sources
Product record checked
2026-07-27
Compare harnessesInspect sources

At a glance

A proprietary terminal surface for Cursor Agent with interactive and print modes, subscription model selection, MCP, project rules, structured output, and allow or deny policies for shell commands and file access.

Good choice if

  • Existing Cursor subscribers who want the same agent from a terminal
  • Scripted analysis and code changes using print mode and structured output
  • Teams that want repository-scoped allow and deny rules without a larger platform

Check before choosing

  • The closed binary has no public immutable release tag or source revision; capability verification depends on dated product documentation and the installed version
  • Model access normally routes through Cursor; the documented Bedrock path does not make the CLI a general local-model harness
  • Sandboxing and worktrees are separate opt-in controls: worktrees isolate files only, while Run Everything/--yolo can bypass interactive approvals
See 3 more considerations
  • Global MCP servers are auto-approved whereas project servers require trust; every extension expands the effective tool boundary
  • Rewind restores file and conversation state but cannot undo network or other external side effects
  • No public product evaluation suite or complete harness benchmark record is available

Why it qualifies as a coding harness

This confirms category fit, not product quality. Every required criterion links back to first-party evidence.

Qualifies4 of 4 required criteria evidenced

  • Adaptive agent loop

    Documented

    The system repeatedly observes results and chooses the next action instead of following a fixed one-pass graph.

    EvidenceCursor CLI overview

  • Repository tool execution

    Documented

    The system can use tools to inspect and change a repository or its execution environment.

    EvidenceHeadless mode

  • Task-aware context management

    Documented

    The runtime assembles, updates, compacts, retrieves, or persists task-relevant context while work proceeds.

    EvidenceUsing Agent in the CLI

  • Model-independent runtime control

    Documented

    Permissions, budgets, interruption, policy, or stop controls operate outside the model's own text generation.

    EvidenceCLI permissions

Membership establishes category fit only. It does not score quality, safety, autonomy, model capability, or benchmark performance. · Read the membership rule.

How it works under the hood

Seven mechanisms mapped from first-party records. These labels describe what the harness provides, not how intelligent its model is.

7/7layers documented
  • Execution & isolationSandbox availableDocumented mechanism, not a performance score.
  • Tooling & integrationsExtensible toolsDocumented mechanism, not a performance score.
  • Context & stateManaged contextDocumented mechanism, not a performance score.
  • Lifecycle & recoveryCheckpoint/rewindDocumented mechanism, not a performance score.
  • ObservabilityLogs/transcriptsDocumented mechanism, not a performance score.
  • VerificationTool-assistedDocumented mechanism, not a performance score.
  • Governance & permissionsPolicy controlsDocumented mechanism, not a performance score.

Public code audit

Unrankedsupport-only repository
Security policy
Present at inspected commit
CI workflow
Not found
Automated tests
Not found
Evaluation assets
Not found
Contributor documentation
Not found

The pinned public tree contains only five paths: a README, SECURITY.md, and an issue-template directory. It exposes no Cursor CLI implementation, release tag, engineering test, CI workflow, contributor guide, or product evaluation suite, so it cannot establish code-verifiable harness capability.

Inspect commit 654b1b4775ca67aef473bd31a14c8c04a1abde2d, checked 2026-07-27

Capability support

Documented first-class product support, checked against the sources below.

External tools (MCP)
DocumentedProduct-supported MCP integrationSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
Local models
Not documentedNo first-class support established by the current recordAbsence of current documentation is not proof that the capability is impossible.
Agent parallelism
DocumentedDelegated or parallel agent workflowSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
Runs without an open UI
DocumentedNon-interactive or automation surfaceSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
Browser control
Not documentedNo first-class support established by the current recordAbsence of current documentation is not proof that the capability is impossible.
Isolated execution
OptionalOptional sandbox policy for command executionSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
Undo file changes
DocumentedProduct-supported file or session rollbackSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.

Primary evidence

Each capability claim is tied to a first-party record and a verification date.

Product record checked 2026-07-27
Cursor CLI overviewInteractive agent, model selection, rules, MCP, installation, subscription access, and product scopeOfficial docsHeadless modePrint mode, structured output, scripting, non-interactive file changes, trust, force behavior, and automationOfficial docsParameters and isolation controlsSandbox, worktree, trust, yolo, MCP, session resume, headless output, ACP, and worker parametersOfficial docsCLI permissionsShell, read, write, web-fetch, and MCP allowlists and denylists plus project and global scopesOfficial docsCLI configurationApproval modes, sandbox and network settings, rewind, workspace trust, web search, and config precedenceOfficial docsCLI changelogDated sandbox, worktree, subagent, checkpoint, rewind, MCP, trust, and auto-review changes through July 2026Official announcementAgent Client ProtocolEditor integration, session protocol, model and mode selection, tools, MCP, and streaming eventsOfficial docsGitHub ActionsCI authentication, headless execution, secrets, permissions, event triggers, and review automationOfficial docs
View 7 additional sources
Product and interfaces3 sources
Cursor Agent CLI announcementRelationship to Cursor Agent, initial headless scope, subscription models, approvals, and risk noticeOfficial announcementCLI installationSupported macOS, Linux, and Windows installation paths, version verification, upgrades, uninstallation, and PATH setupOfficial docsUsing Agent in the CLIInteractive prompting, rules, MCP use, conversation navigation, change review, command approval, and session workflowsOfficial docs
Agents, state and recovery1 sources
CLI Shell ModeContinuous shell interaction, retained conversational context, command history, and long-running development sessionsOfficial docs
Automation and extensions1 sources
CLI MCPModel Context Protocol server configuration, tool discovery, connection management, and CLI invocationOfficial docs
Enterprise and operations1 sources
CLI authenticationInteractive login and API-key authentication used by local, headless, and continuous-integration CLI workflowsOfficial docs
Releases and public code audit1 sources
Cursor public support repository snapshotPinned README, security policy, and issue template surface; it does not expose Cursor CLI source, tests, releases, or evaluationsOfficial repository