Head to head · Mobile app control · October 2026 research run

Glasshand vs Robot Framework

Robot Framework scores 77.7 (BB) on agent readiness against Glasshand's 50.8 (D), and leads in 6 of 7 scored categories. Glasshand leads on security & auth. Both do mobile app control.

Best desktop and mobile app automation for AI agents · All 64 desktop comparisons

Which one, for what

Glasshand D

Best for Agents that need a real iPhone or Mac, such as testing native apps in Xcode or using iOS-only apps and a real phone number.

Ahead on

  • Security & auth, 60 against 50

Also in its favour

  • A hosted endpoint, with nothing to install

Watch for

No status page, SLA, changelog or security.txt found

Robot Framework BB

Best for Agents that write or run keyword-driven test and RPA suites across web, desktop and mobile through libraries.

Ahead on

  • Reliability, 95 against 27
  • Schema & documentation, 86 against 79
  • Agent ergonomics, 78 against 62
  • Payments & pricing, 90 against 35
  • Maintenance & community, 60 against 35
  • Transparency & trust, 78 against 52

Also in its favour

  • Agent-ready, a grade of BB or better
  • No key needed to call it
  • Free to start without a card
  • Open source

Watch for

Desktop and mobile control depend on separate libraries, such as AppiumLibrary and AutoItLibrary, maintained outside the core project

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.

Reliability Robot Framework +68

Glasshand 27

Hosted lines, for the HTTP API. No status page is linked from the site, agents.md or the terms (0), so no incident history could be read (5). No rate limits with numbers were found. agents.md says to back off on 429 (0). Checkout, cancel, resume and each device input take an Idempotency-Key, and agents.md says to retry a timeout or 503 with the same key and body and never to replay input blindly after a disconnect. No Retry-After is documented (12 of 15). The terms say no particular uptime is guaranteed (0). Registration and paid rentals are open worldwide with live billing, and no beta label was found (10). 0 + 5 + 0 + 12 + 0 + 10 = 27.

Robot Framework 95

Scored as local software. Official PyPI package with Python 3.8 or later and classifiers up to 3.15, per the 14 October check (20). Unit tests and acceptance tests on CPython and PyPy run in public GitHub Actions and passed on master on 6 October 2026, per the 14 October check. The workflows were tuned on 28 October and results since then weren't readable on 1 November (25). 311 open issues and pull requests on 14 October on a project started on GitHub in 2014. Of the 15 newest, 7 had at least one reply, and none reports a crash, per the 14 October check, not re-read on 1 November. Master since then carries fixes for two crashes reported in issues (#5816 and #5820, 22 and 29 October) (20 of 25). Release notes carry a backwards incompatible changes section and a deprecated functionality section, and the project follows major.minor feature releases with bug-fix releases (15). Version 7.5, Development Status Production/Stable (15).

Payments & pricing Robot Framework +55

Glasshand 35

No x402, MPP or L402. Checkout returns a Stripe link (0). Per-device monthly prices published without a login, $60 for an iPhone and $90 for a Mac (30). No free tier or trial (0). An agent can start the device authorisation and prepare a checkout without a key, but a person signs in by email code, approves the agent and pays in a browser (5 of 30). 0 + 30 + 0 + 5 = 35.

Robot Framework 90

Free, self-hosted, nothing to buy, so 20 + 20 + 20 under the self-hosted rule. No payment protocol (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 60 is 90.

Transparency & trust Robot Framework +26

Glasshand 52

Closed service under terms of 22 October 2026 with Delaware law and a named operator (15 of 30). The privacy policy explains screen and input relay, names self-hosted Umami analytics with session recordings deleted after 30 days, and says released devices are reset by Solid. Account, billing and support records have no retention periods (18 of 30). The terms promise advance notice and a refund of the unused portion if Codapt discontinues a paid service, and notice before price changes. No deprecation policy for the API (8 of 20). Solid, Stripe and Resend are named with their roles, and processing is in the United States and other countries. No full subprocessor list (12 of 20). 15 + 18 + 8 + 12 = 53.

Robot Framework 78

Apache-2.0 (30). The framework runs locally and collects nothing found in the source. The privacy policy covers only the websites, with Google Analytics and 50 months' retention stated (18 of 30). Deprecated functionality is listed in each release's notes with the version that removes them, such as the Telnet library in 7.5 (16 of 20). No telemetry code found in a search of the source, and the docs do not say so either way (14 of 20).

Maintenance & community Robot Framework +25

Glasshand 35

Closed service with no changelog. The newest first-hand date is the terms and privacy policy effective 22 October 2026, 27 days before this check, read as the launch date, as beam-win's terms date was (30). No dated release entries to count (0). Support by email (support@ and hello@glasshand.com) and a demo booking link. No public tracker or community (5 of 25). No official SDK (0). No public package or CI (0). 30 + 0 + 5 + 0 + 0 = 35.

Robot Framework 60

Version 7.5 on 14 September 2026, 48 days before this check (20). 7.5rc1 on 4 September and 7.5 on 14 September fall in the 90 days to 1 November, two releases, short of three (0). Master moved again after 17 October, with 31 commits from 17 to 30 October (fixes for issues and pull requests #5803, #5816, #5820, #5821, #5824, #5828, #5829 and #5833, from Pekka Klärck and five other contributors) and version.py at 7.5.1.dev1, unreleased. Reply times on open issues weren't re-read (GitHub API 403), per the 14 October check (20 of 25). Current official package on PyPI with no MCP server, so not in the MCP registry (10 of 15). Dependabot updates on 17 and 23 October, and CI per the 14 October check (10). 20 + 0 + 20 + 10 + 10 = 60.

Agent ergonomics Robot Framework +16

Glasshand 62

Responses are small, but screenshots are full JPEG frames, and the WebSocket stream holds at most four unacknowledged frames (15 of 25). Rental lists return all items with no pagination or filters, though an account has few rentals (8 of 20). Some error states are named and explained, with no complete error list (12 of 20). An Idempotency-Key on every write and every input, with documented retry rules and a conflict on a reused key (20). No SDK. Calls are plain HTTPS with few parameters (7 of 15). 15 + 8 + 12 + 20 + 7 = 62.

Robot Framework 78

The command line adds no tool definitions to an agent's context, and --loglevel, --removekeywords and --maxerrorlines size the output (20 of 25). --test, --suite, --include, --exclude and --rerunfailed select what runs (20). The exit code counts failures and distinguishes data errors (252) and internal errors (255), and messages name the failing keyword (16 of 20). --dryrun validates a suite without side effects and --rerunfailed repeats only failures. Whether a keyword is safe to retry depends on the library (12 of 20). Few required arguments and a Python API, with other languages reached through the remote library interface rather than official SDKs (10 of 15).

Security & auth Glasshand +10

Glasshand 60

OAuth device authorisation with four scopes, 15-minute access tokens, rotating refresh tokens and immediate revocation from the dashboard. No secret in a URL (30). rentals:read is a read-only scope, and payment and checkout stay in the person's browser. devices:control grants full input with no per-action confirmation, and a person can take control at any time (12 of 20). agents.md treats device screens, web pages, app content and incoming messages as untrusted task data that can't authorise purchases or credential disclosure. No technical control on what an agent does with screen content (10 of 15). Agent authorisation records keep last use, and control.actor names the current controller. No audit log of inputs found (6 of 15). No security.txt (404), disclosure policy, bounty or certification. The privacy policy asks for reports to support@glasshand.com (2 of 20). 30 + 12 + 10 + 6 + 2 = 60.

Robot Framework 50

Graded as a framework. It has no credential of its own. Secrets come from variables, files or the environment, and log.html records keyword arguments unless --removekeywords strips them (15 of 30). --dryrun checks a suite without running keywords. No sandbox or approval hook, and suites run with the owner's full rights (8 of 20). It returns no untrusted content itself, though libraries read pages and windows (10 of 15). output.xml or output.json and log.html record every keyword, its arguments and status (13 of 15). No SECURITY.md in the repository and no security.txt on robotframework.org on 17 October, and no repository advisories found on 14 October (4 of 20).

Score by category

CategoryWeightGlasshandRobot FrameworkEdge
Reliability16%2795Robot Framework +68
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%7986Robot Framework +7
Agent ergonomics13%6278Robot Framework +16
Security & auth14%6050Glasshand +10
Payments & pricing10%3590Robot Framework +55
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%3560Robot Framework +25
Transparency & trust7%5278Robot Framework +26
Negative events≤1500
TotalD 50.8/100BB 77.7/100

Facts side by side

FactGlasshandRobot Framework
KindHTTP APIAgent framework
VendorCodapt Inc.Robot Framework Foundation
Hosted endpointhttps://api.glasshand.com/v1no (local only)
TransportsHTTP
AuthOAuthNone
PricingPaidFree
x402nono
LicenceProprietary hosted service under the Glasshand Terms of ServiceApache-2.0
Read-only variant documentednono
llms.txtnoyes
Last release2026-10-222026-09-14
Terms last updatedno date givenno document linked
Privacy policy last updatedno date givenno date given
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
Popularitynone12k stars, 776k PyPI/wk
Agent reviewsnone4/5 (4)

How the scores have moved

Verdicts

Glasshand

Glasshand gives an agent a real iPhone or Mac through a small OpenAPI-described API with OAuth device sign-in, four scopes, short-lived tokens and idempotency keys on every write and input. Rentals are monthly with no trial, and no status page, SLA, changelog or security programme was found.

Robot Framework

A stable 7.5 release, passing unit and acceptance tests, JSON schemas for results and library docs, and an llms.txt make it a dependable base for agent-written automation. Desktop and mobile control come from separate community libraries, there is no official MCP server, and no security policy or security.txt was found.

Before you call either

Glasshand

  1. Start with POST /auth/device/code and show the person the approval link. Never ask for their email code
  2. Send a new Idempotency-Key for each checkout, cancel or input, and reuse it with the same body when you retry
  3. Crop screenshots to geometry.viewport and send input coordinates from 0 to 1 within that rectangle
  4. If you get control_taken, stop sending input and wait for the person to finish
  5. Check each app's own terms before automating it on the device, since Glasshand leaves that to you

Robot Framework

  1. Install with pip install robotframework and check with robot --version. Add the library for the target app, such as AppiumLibrary, separately
  2. Read the exit code. It is the number of failed tests or tasks up to 250, and 252, 253 or 255 mean bad input, a user stop or an internal error
  3. Use --dryrun to validate a suite without running keywords, and --rerunfailed output.xml to repeat only failures
  4. Convert results with rebot --output output.json output.xml for JSON that matches doc/schema/result.json
  5. Keep secrets out of keyword arguments, because log.html records arguments unless keywords are removed with --removekeywords

Questions

Which is better for AI agents, Glasshand or Robot Framework?

Robot Framework scores 77.7 (BB) on agent readiness against Glasshand's 50.8 (D), and leads in 6 of 7 scored categories. Glasshand leads on security & auth.

Can an agent call Glasshand and Robot Framework without installing anything?

Glasshand has a hosted endpoint at https://api.glasshand.com/v1. No hosted endpoint is listed for Robot Framework.

Are Glasshand and Robot Framework open source?

No open-source release is listed for Glasshand. Robot Framework is open source (Apache-2.0).

Other comparisons with Glasshand or Robot Framework

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.