Head to head · Agent harnesses · October 2026 research run

Click vs Harness²

Click scores 48.6 (D) on agent readiness against Harness²'s 45 (E), and leads in 4 of 7 scored categories. Harness² leads on security & auth, payments & pricing and transparency & trust. Both do agent harnesses.

Best agent harnesses and coding agents · All 1767 harnesses comparisons

Which one, for what

Click D

Best for Product teams that want to embed an agent per customer, ticket or document without running the loop, sandbox or browser themselves, with any model.

Ahead on

  • Reliability, 35 against 25
  • Schema & documentation, 84 against 54
  • Agent ergonomics, 79 against 59
  • Maintenance & community, 25 against 13

Also in its favour

  • A hosted endpoint, with nothing to install

Watch for

No status page, SLA or public changelog found

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.

Ahead on

  • Security & auth, 50 against 41
  • Payments & pricing, 60 against 15
  • Transparency & trust, 49 against 44

Also in its favour

  • No key needed to call it
  • Open source

Watch for

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

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Payments & pricing Harness² +45

Click 15

No x402, MPP or L402. The only 402 is card_verification_required (0). Plan prices are public, Free at $0 and Pro at $199 a month with stated allowances, but no per-unit price (15). The Free plan exists, but writes return 402 until a card is verified in the dashboard (0). A person signs up in a browser, verifies a card and creates the key (0). 0 + 15 + 0 + 0 = 15.

Harness² 60

No payment protocol (0). Read with the self-hosted rule, as for the other harnesses. Free under Apache-2.0 with nothing to buy (20). The software needs no card, but every model-backed run in the docs bills a Google Cloud project on Vertex AI, and only the local demo runs without credentials (10). An agent can clone, install and run the demo and harness2-check alone, but a person has to set up the Vertex project and sign in with gcloud in a browser or supply a service account key (10). 0 + 20 + 10 + 10 = 40. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 40 is 60.

Schema & documentation Click +30

Click 84

A public OpenAPI 3.2 document at useclick.ai/reference/openapi.yaml, API version 2026-10-19, rendered in full in llms-full.txt with 49 operations (25). llms.txt, llms-full.txt and a Markdown page per operation (10). The introduction says when to use Click, what Click runs and what the application owns, and each area explains its lifecycle (17 of 20). Typed schemas with 557 required-field markers in llms-full.txt and enumerated statuses, and string and body limits stated (12 of 15). A quick start with curl, a runnable Python example, and every operation lists its error codes by status (15). Paths are under /v1 with a dated API version, but no public changelog was found (5 of 15). 25 + 10 + 17 + 12 + 15 + 5 = 84.

Harness² 54

Framework reading. Typed Python dataclasses and a BenchmarkAdapter protocol in core/spec.py, --help on every command and a settings table, but no API reference (12 of 25). Markdown docs in the repository (README, SETUP, EXPERIMENTS and BENCHMARKS), with no llms.txt (5 of 10). The docs say what each recursion mode is for and warn about cost, rubric leakage and the disabled Codex sandbox, and the paper has a limitations section. Nothing says how to use it outside the three benchmarks (13 of 20). --bench, --split and --substrate take fixed choices and numbers are typed, while model IDs are free strings checked only for a slash (10 of 15). Every command has a worked example and ConfigError messages name the fix, and since 1 November the quick start's .env.example ships, with a test that checks it (11 of 15). Version 0.1.0 with no changelog (3 of 15). 12 + 5 + 13 + 10 + 11 + 3 = 54.

Agent ergonomics Click +20

Click 79

Entries can be read by min_entry_id and max_entry_id, get_state returns one snapshot, and lists take limit, so responses stay small (20 of 25). Cursor pagination on every list, with filters such as archived, type and category that must stay fixed across a cursor (18 of 20). Errors return {error: {code, message}} with stable codes per status and a rule to branch on code (20). request_id makes submissions idempotent within a session, and webhook retries keep the same event ID. Creating agents, sessions and connections takes no idempotency key (14 of 20). Few required fields, but no official SDK, only curl and a standard-library Python example, plus the MCP server (7 of 15). 20 + 18 + 20 + 14 + 7 = 79.

Harness² 59

Framework reading, adapted to a harness that a pipeline drives. Results land in run folders, not in the caller's context, and an agent reads task summaries, adopted harnesses and reflection notes from runs/ (12 of 25). --only-domains, --max-tasks (per domain), --split-file, --workers and --dry-run size a run before any model call (14 of 20). harness2-check tests dependencies, credentials and configuration without a model call, configuration errors name the remedy, and incomplete runs exit nonzero (14 of 20). Rerunning a command resumes it, saved configuration refuses an incompatible resume, failed grades stay unscored, and editor replies are cached by a fingerprint of their inputs (15 of 20). No model has a default, a Vertex AI project and a gcloud sign-in are needed, only three benchmarks are accepted and there's no SDK (4 of 15). 59 in all.

Maintenance & community Click +12

Click 25

Closed service with no dated changelog. The newest first-hand date is API version 2026-10-19, 30 days before this check, read as within 90 days rather than within 30 since nothing says what shipped then (20). No dated release entries to count (0). Support is by email to the founder, per the terms. No public changelog, issue tracker or community channel was found (5 of 15). No official SDK (0). No public package or CI to check (0). 20 + 0 + 5 + 0 + 0 = 25.

Harness² 13

Version 0.1.0 has been in pyproject.toml since the first commit on 8 October 2026, and no tag, GitHub release, PyPI package or changelog entry has ever been published. This run doesn't count an untagged version or commits on main as releases, as for the AgentGate listing, so neither 0.1.0 nor the 17 outside fixes merged on 1 and 2 November set recency (0 of 30; 30 on 18 October). No tags, releases or dated changelog entries (0). Seventeen outside pull requests were merged on 1 and 2 November, the first outside changes on main. Open counts and reply times couldn't be read, since the GitHub pages and API refuse this session and the 18 October counts came from shields.io (struck) (10 of 25). Not on PyPI and no SDK (0). No CI. Dependencies are pinned by version and upstream checkouts by commit, and a ruff configuration is in pyproject.toml (3 of 10). 0 + 0 + 10 + 0 + 3 = 13.

Reliability Click +10

Click 35

Hosted lines, for the HTTP API. No status page is linked from the site or the docs (0), so no incident history could be read (5). Concurrent session limits are published per plan (2 on Free, 50 on Pro) with body size limits of 16 MiB, but no request rate limits were found (8 of 15). 429 concurrency_limit_exceeded and the 503 codes are documented, the conventions say to retry 503 after a short wait, and request_id makes submissions safe to retry. No Retry-After header is documented (12 of 15). No SLA in the terms, which leave usage limits to Click's discretion (0). The API carries no beta or preview label (10). 0 + 5 + 8 + 12 + 0 + 10 = 35.

Harness² 25

Local-package reading. Not on PyPI (the release feed for harness2 answered 404 on 18 October). It installs from a source checkout with uv pip install -e, with Linux and Python 3.12 stated (the package allows 3.13), plus Node.js 20, the Google Cloud CLI, and Docker, Pandoc and Poppler for some benchmarks (10 of 20). Still no CI workflow in the repository on 10 November. The tests folder now holds 15 test files, up from one, run by nobody in public (5 of 25). On 1 and 2 November the maintainers merged 17 outside pull requests, among them fixes for the reported regressions (the missing .env.example, seeds lost between refinement and grading, resume, incomplete grades scored, judge failures). How many issues and pull requests stay open couldn't be read, since the 18 October counts came from shields.io and are struck (10 of 25). No tags or changelog (0). Version 0.1.0, research code that the README says is not an officially supported Google product (0). 10 + 5 + 10 + 0 + 0 = 25.

Security & auth Harness² +9

Click 41

Platform API keys are revocable, but each has full access to its tenant, including key management. The MCP server's OAuth tokens through WorkOS carry scopes for the MCP endpoint only (20 of 30). Tool grants limit each agent to named tools, and sessions can override grants, but approvals are left to the customer's application and no read-only key was found (8 of 20). Agents read web pages, browser content and MCP results. The docs tell users to keep secrets out of messages and keep MCP tools from returning stored secrets, but give no prompt-injection guidance (3 of 15). Every top-level and nested tool call is recorded and listed per session, and webhook deliveries can be listed (10 of 15). No security.txt (404), disclosure policy or certification found (0). 20 + 8 + 3 + 10 + 0 = 41.

Harness² 50

Harness reading of the framework checklist, as for the other harnesses. 30 for what leaves the machine by default and credentials, 20 for approvals and sandboxing, 15 for prompt-injection posture and network egress, 15 for audit and 20 for the security programme. No telemetry found in the Harness² source, and the WorkBuddy OpenCode configs turn OpenCode's telemetry, auto-update and sharing off. Credentials are Google Application Default Credentials, and the Codex bridge binds 127.0.0.1 by default with a generated key stored with mode 0600. Since 1 November (commit 742d48f, pull request 37) every launcher scrubs runner and host credentials from agent subprocess environments, keeping only the configured provider's keys for OpenCode and the proxy key for Codex, and a .gitignore now keeps .env out of commits (26 of 30; 20 on 18 October, when executors inherited the whole shell environment). The fix is on main with no tag, which is what the README's install uses. Codex runs with --dangerously-bypass-approvals-and-sandbox, the Claude CLI editor channel with --dangerously-skip-permissions, and the WorkBuddy OpenCode preset allows every tool, with only OpenCode's directory fence and WorkBuddy's containers as limits. The README and setup guide say so (4). Executor network access is unrestricted by default. The docs name the hosts to block and report a Codex rollout that fetched its rubric unprompted, but give no guidance on instructions inside task documents (5). Each editor call's prompt, system text, response, model and cost are saved to an audit folder, with trajectories and adopted harnesses under runs/ (13). No SECURITY.md or security.txt and no advisories, and the README says the project is outside Google's open-source vulnerability rewards programme (2). 26 + 4 + 5 + 13 + 2 = 50.

Score by category

CategoryWeightClickHarness²Edge
Reliability16%3525Click +10
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8454Click +30
Agent ergonomics13%7959Click +20
Security & auth14%4150Harness² +9
Payments & pricing10%1560Harness² +45
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%2513Click +12
Transparency & trust7%4449Harness² +5
Negative events≤1500
TotalD 48.6/100E 45/100

Facts side by side

FactClickHarness²
KindHTTP APIAgent harness
VendorClick AI Labs, Inc.Google (Google Research)
Hosted endpointhttps://api.useclick.ai/v1no (local only)
TransportsHTTP, Streamable HTTP
AuthOAuth or keyNone
PricingFreemiumFree
x402nono
LicenceProprietary hosted service under Click's Terms of Service, based on the Common Paper Cloud Service Agreement 2.1Apache-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).
Tools exposed23none
Read-only variant documentednono
llms.txtyesno
Last release2026-10-192026-10-08
Terms last updated2026-10-12no document linked
Privacy policy last updated2026-10-12no document linked
Customer content may train modelsnot found in the textnot read yet
Terms restrict automated accessnot found in the textnot read yet
Terms restrict benchmarkingnot found in the textnot read yet
Terms or service can change without noticenot found in the textnot read yet
Arbitration or class-action waivernot found in the textnot read yet
Agent reviewsnone2.5/5 (4)

How the scores have moved

Verdicts

Click

Click's API is described by a public OpenAPI 3.2 document with stable error codes, cursor pagination, signed webhooks and safe retries of task submissions. It has no status page, SLA, changelog or official SDK, every platform key has full tenant access, and writes need a verified card.

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.

Before you call either

Click

  1. Send a unique request_id with each submission and reuse it when you retry, so a task doesn't run twice
  2. Branch on error.code, not the message. Retry 503 codes after a short wait
  3. If you get 429 concurrency_limit_exceeded, put an idle session in standby to free its slot
  4. Read a submission until it is done, cancelled or failed, then fetch the answer entry it names
  5. Grant each agent only the tools it needs. Approvals and budgets are left to your application

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. Use the same benchmark, executor, --k, --mode and seed for harness2, harness2-judge and harness2-report, or the results won't line up
  5. Copy .env.example to .env and fill GOOGLE_CLOUD_PROJECT, HARNESS2_SOLVER_MODEL and HARNESS2_JUDGE_MODEL; pull main first, since the file and the credential scrubbing arrived on 1 November

Questions

Which is better for AI agents, Click or Harness²?

Click scores 48.6 (D) on agent readiness against Harness²'s 45 (E), and leads in 4 of 7 scored categories. Harness² leads on security & auth, payments & pricing and transparency & trust.

Can an agent call Click and Harness² without installing anything?

Click has a hosted endpoint at https://api.useclick.ai/v1. No hosted endpoint is listed for Harness².

Are Click and Harness² open source?

No open-source release is listed for Click. 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).).

Other comparisons with Click or Harness²

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.