HarnessMatch
Menu

Coding agent

Kiro CLI

AWS-backed terminal agent with governed tools, parallel subagents, goals, and CI mode.

Status
Active
License
Proprietary
Evidence
Documented, 21 sources
Product record checked
2026-07-27
Compare harnessesInspect sources

At a glance

A proprietary coding CLI with interactive and headless execution, custom and parallel agents, MCP, steering, hooks, resumable sessions, goal loops, and an optional classic checkpoint system; CLI 3.0 adds a separate early-access harness and policy format.

Good choice if

  • AWS-aligned teams that need governed terminal and CI automation
  • Parallel research or implementation with permission-scoped subagents and review loops
  • Longer interactive tasks that benefit from goals, steering, session resume, and optional file checkpoints

Check before choosing

  • Commands and tools operate in the local environment; no built-in execution sandbox or local open-weight model path was verified, and Supervised versus Autopilot changes review behavior rather than capability, isolation, or access control
  • Headless runs require an API key and pre-trusted tool categories, so least-privilege configuration is part of the CI setup
  • Classic checkpoints are experimental and session-scoped; they do not undo external side effects
See 2 more considerations
  • CLI 3.0 is early access, opt-in, changes the permission and session formats, removes supervised mode, and cannot use the classic non-TUI path; its declarative permission rules are therefore not treated as the stable 2.x default
  • The goal loop's completion check is agent-driven vendor functionality, not independent evidence that acceptance criteria were actually satisfied

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.

    EvidenceGoal loop

  • Repository tool execution

    Documented

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

    EvidenceBuilt-in tools

  • Task-aware context management

    Documented

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

    EvidenceSession management

  • Model-independent runtime control

    Documented

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

    EvidenceTool 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 & isolationHost processDocumented 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
Not found
CI workflow
Present at inspected commit
Automated tests
Present at inspected commit
Evaluation assets
Not found
Contributor documentation
Present at inspected commit

The public repository contains product guidance, issue templates, issue-triage automation, and tests for those support workflows. It does not expose the proprietary Kiro IDE or CLI implementation, agent harness, checkpoint engine, or product evaluation suite.

Inspect commit e8daa058590dd58efb14f6d41ddb3ba1a26cfba3, checked 2026-07-27

Measured configurations

No benchmark run passes the full metadata admission policy for this harness yet. Missing data is not scored as zero.

Benchmark policy and all runs

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
Not documentedNo first-class support established by the current recordAbsence of current documentation is not proof that the capability is impossible.
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
Kiro CLI 2.14.0 releaseCurrent release, interrupted-response retries, and opt-in V2-to-V3 agent migrationOfficial announcementCLI quickstartTerminal workflow, plan mode, approval prompts, project steering, and MCP onboardingOfficial docsSubagentsUp to four parallel agents, task graphs, review loops, live monitoring, aggregation, and scoped tool permissionsOfficial docsTool permissionsPer-tool trust, command and filesystem approvals, custom-agent permissions, and headless trust behaviorOfficial docsHeadless modeAPI-key authentication, non-interactive execution, scoped tool trust, MCP startup gates, CI examples, and limitationsOfficial docsCLI commands and sessionsInteractive and non-interactive flags, session resume, JSON-formatted settings, diagnostics, and log controlsOfficial docsSession managementPer-turn autosave, per-directory resume, export and import, SQLite storage, and custom persistence scriptsOfficial docsConversation rewindNon-destructive conversation branching with tool-call and context previews; not file rollbackOfficial docs
View 13 additional sources
Execution and control3 sources
Built-in toolsFile, shell, search, goal, web, subagent, and other built-in tool semantics and settingsOfficial docsPrivacy and security modelLocal execution boundary and explicit distinction between Supervised or Autopilot review behavior and sandboxing, isolation, or access controlOfficial docsCLI 3.0 permission rulesEarly-access declarative allow, ask, and deny capability rules, precedence, matching, project scope, and auditabilityOfficial docs
Agents, state and recovery4 sources
Goal loopIterative implementation and agent-driven verification, acceptance criteria, iteration limits, steering, and cancellationOfficial docsQueue steeringMid-turn redirection at tool boundaries, queued follow-ups, cancellation, and timing limitationsOfficial docsClassic checkpointingOpt-in shadow-Git snapshots, per-turn and per-tool diffs, soft and hard restore, session scope, and cleanupOfficial docsCustom agentsAgent-specific prompts, tools, resources, permissions, hooks, MCP servers, and configuration scopeOfficial docs
Automation and extensions2 sources
HooksLifecycle and tool hooks, JSON events, blocking pre-tool decisions, logging, formatting, and validationOfficial docsMCP integrationUser and workspace MCP servers, transport, OAuth, tool loading, trust, and enterprise registry controlsOfficial docs
Enterprise and operations2 sources
Settings referenceCheckpoint feature flag, compaction, logs, tool search, Kiro home isolation, and version-specific settingsOfficial docsData protectionEncryption, individual and enterprise service-improvement treatment, telemetry, and opt-out controlsOfficial docs
Releases and public code audit2 sources
CLI 3.0 early accessOpt-in unified harness, capability policies, specs, breaking changes, session incompatibility, and known gapsOfficial docsCLI 2.7 goal and steering releaseIntroduction of goal loops, queue steering, and enriched rewind previewsOfficial announcement