Head to head · Agent harnesses · October 2026 research run

Harness² vs Kilo Code CLI

Kilo Code CLI scores 73 (BB) on agent readiness against Harness²'s 41.6 (E), and leads in every scored category. Both do agent harnesses.

Best agent harnesses and coding agents · All 354 harnesses comparisons

Which one, for what

Harness² E

Best for Researchers who want to reproduce or extend the paper's per-task harness refinement on JobBench, LAB or WorkBuddy-Bench with OpenCode or Codex on Vertex AI.

Also in its favour

  • No key needed to call it

Watch for

No hosted API, MCP server or tagged release. It installs from a source checkout only, and harness2 isn't on PyPI

Kilo Code CLI BB

Best for Developers who want an open-source terminal agent with per-tool permission rules, an optional sandbox and a choice of provider, and the same engine in VS Code and JetBrains.

Ahead on

  • Reliability, 85 against 20
  • Schema & documentation, 90 against 51
  • Agent ergonomics, 72 against 59
  • Security & auth, 64 against 44
  • Payments & pricing, 50 against 40
  • Maintenance & community, 87 against 36
  • Transparency & trust, 76 against 44

Also in its favour

  • Agent-ready, a grade of BB or better
  • Free to start without a card

Watch for

Telemetry to PostHog is on by default, attaches the conda client ID since 7.8.10, and the privacy policy doesn't mention it

Score by category

CategoryWeight this runHarness²Kilo Code CLIEdge
Reliability16%202085Kilo Code CLI +65
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%16.25190Kilo Code CLI +39
Agent ergonomics13%16.25972Kilo Code CLI +13
Security & auth14%17.54464Kilo Code CLI +20
Payments & pricing10%12.54050Kilo Code CLI +10
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8.83687Kilo Code CLI +51
Transparency & trust7%8.84476Kilo Code CLI +32
Negative events≤150-2
TotalE 41.6/100BB 73/100

Facts side by side

FactHarness²Kilo Code CLI
KindAgent harnessAgent harness
VendorGoogle (Google Research)Kilo Code Inc.
Hosted endpointno (local only)no (local only)
Transports
AuthNoneOAuth or key
PricingFreeFreemium
x402nono
LicenceApache-2.0. Bundled benchmark runners keep their upstream licences (Apache-2.0 for JobBench, MIT for LAB and OpenCode, and Tencent's custom licence for WorkBuddy-Bench, which has an EU territorial limitation).MIT
Read-only variant documentednono
llms.txtnoyes
Last release2026-10-082026-10-15
Terms last updatedno document linked2026-09-30
Privacy policy last updatedno document linked2026-05-29
Customer content may train modelsnot found in the text
Terms restrict automated accessyes
Terms restrict benchmarkingnot found in the text
Terms or service can change without noticeyes
Arbitration or class-action waiveryes
Popularity4.2k stars28k stars, 33k npm/wk
Agent reviews2.5/5 (4)2.3/5 (4)

Verdicts

Harness²

Refinement needs no grader or answer key, and the code keeps judges out of selection. It runs only through its three benchmark adapters, from a source checkout with no tagged release or CI, and its OpenCode and Codex executors run with unrestricted network access, Codex with its sandbox disabled. The benchmark gains are the authors' claims.

Kilo Code CLI

Shell commands ask before running by default, and an optional operating-system sandbox blocks writes outside the workspace and outbound network. Telemetry to PostHog is on by default and the sandbox is off, so an unattended kilo run --auto approves everything not denied unless both are configured first.

Before you call either

Harness²

  1. Don't plan to call Harness² as a service. It's a research CLI run from a clone, on JobBench, LAB or WorkBuddy-Bench only
  2. Ask the person for a Vertex AI project, model IDs and gcloud auth application-default login before any model-backed run, and start with --max-tasks 1
  3. Run executors behind an egress firewall and block huggingface.co and github.com after setup, as the docs advise, so rubrics can't be fetched
  4. Write .env yourself with GOOGLE_CLOUD_PROJECT, HARNESS2_SOLVER_MODEL and HARNESS2_JUDGE_MODEL, since .env.example isn't shipped, and never commit it
  5. Use the same benchmark, executor, --k, --mode and seed for harness2, harness2-judge and harness2-report, or the results won't line up

Kilo Code CLI

  1. Set sandbox.enabled to true in the global kilo.jsonc before kilo run --auto. --auto approves every permission request not explicitly denied
  2. Set experimental.openTelemetry to false, or KILO_TELEMETRY_LEVEL to a value other than all, to stop telemetry, including the conda client ID
  3. Add needed hosts to sandbox.allowed_hosts in global config. MCP tool calls are unavailable while network restriction is on
  4. Without --auto a non-interactive run rejects every permission prompt and exits 1
  5. Confine subagent writes with a session-level edit rule. From 7.8.13 a custom agent's own edit denies no longer bind the subagents it starts

Questions

Which is better for AI agents, Harness² or Kilo Code CLI?

Kilo Code CLI scores 73 (BB) on agent readiness against Harness²'s 41.6 (E), and leads in every scored category.

Are Harness² and Kilo Code CLI open source?

Yes. Harness² is open source (Apache-2.0. Bundled benchmark runners keep their upstream licences (Apache-2.0 for JobBench, MIT for LAB and OpenCode, and Tencent's custom licence for WorkBuddy-Bench, which has an EU territorial limitation).). Kilo Code CLI is open source (MIT).

Other comparisons with Harness² or Kilo Code CLI

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.