HarnessmarketEnter the desk
/HMX Claude Code 29.1%/Open index 519 slips/Meridian desk 6,400 agents live/Skills + MCP 553,855 indexed (Skillful, Aug 2026)/Agent Skills spec 1.2M+ open packages/Deloitte: orchestration worth +15–30% of autonomous-agent TAM by 2030/Gartner: 40% of agentic projects cancelled by 2027 without a supervisor/SWE-bench Pro: harness swap > many model upgrades/HM-SWE-Pro: Loop+Claude 41.2 pass@1 · chat 18.2 · same model/Lewis 2608.26218: F2PF 28→49 under a tighter harness, same model/ACES: Skill Lift 0.21 · 947 paired cases · scan vs live ρ=0.14/Tokenomics: code review 59.4% of ChatDev tokens/HMX Claude Code 29.1%/Open index 519 slips/Meridian desk 6,400 agents live/Skills + MCP 553,855 indexed (Skillful, Aug 2026)/Agent Skills spec 1.2M+ open packages/Deloitte: orchestration worth +15–30% of autonomous-agent TAM by 2030/Gartner: 40% of agentic projects cancelled by 2027 without a supervisor/SWE-bench Pro: harness swap > many model upgrades/HM-SWE-Pro: Loop+Claude 41.2 pass@1 · chat 18.2 · same model/Lewis 2608.26218: F2PF 28→49 under a tighter harness, same model/ACES: Skill Lift 0.21 · 947 paired cases · scan vs live ρ=0.14/Tokenomics: code review 59.4% of ChatDev tokens

Harness · managed

Claude Code

The productised coding harness. Tool loop, permissions, Skills, production surfaces.

OriginSubscribe on desk

Publisher

Anthropic

License

proprietary

Latest

2.4.1

Installs

184,200

Thesis

Claude Code is the reference commercial harness: a disciplined tool loop, a permission system operators can actually audit, and native Agent Skills. It is the default production install in mixed fleets that standardise on Anthropic.

On SWE-bench Pro, harness choice moved pass@1 more than several model upgrades. Claude Code is the reason that sentence exists. The listing covers the terminal, IDE, desktop, and managed-agent surfaces under one signed tree. Enterprise installs bind the permission manifest to the organisation allowlist; community Skills do not load in production unless restamped certified.

Direction

This listing is a reference. The tree lives at origin. The command is how you load it here.

Subscribe (seat) → hm install anthropic/claude-code@2.4.1

Install via the registry. Pin a version. Bind the org allowlist. Do not run unsigned community Skills in production — that is how the 2026 supply-chain incidents happened.

Versions

SemverSHAReleasedNotesSigned
2.4.1a91c3e22026-08-18Skills directory sync; tighter sandbox on network tools.signed · sbom
2.3.077b1d0a2026-06-02Managed-agent parity with CLI permissions.signed · sbom
2.1.0c0ffee92026-03-11Agent Skills GA. Permission manifests required.signed · sbom

Subscribe

$100/mo

Seat · seat / month

Enterprise from $48,000

Subscribe, then pin. Entitlement is checked at deploy, not in a PDF.

Origin

code.claude.com

via lab

Open origin →

Compatibility

  • claude-code

Permission manifest

fs.read
**
fs.write
**
network
registry.harness.market, api.github.com
exec / secrets
exec · GITHUB_TOKEN

Publisher

Anthropic

Origin of the Agent Skills spec and the Claude Code harness.