Head to head · Browser automation · October 2026 research run

agent-browser vs Tilion Fortress

agent-browser scores 67.9 (B) on agent readiness against Tilion Fortress's 55.9 (C), and leads in every scored category. Both do browser automation.

Best browser automation tools for AI agents · All 308 browser comparisons

Which one, for what

agent-browser B

Best for Coding agents with a shell that want short commands and compact snapshots.

Ahead on

  • Schema & documentation, 84 against 74
  • Agent ergonomics, 80 against 73
  • Security & auth, 67 against 35
  • Payments & pricing, 90 against 60
  • Maintenance & community, 70 against 47
  • Transparency & trust, 63 against 58

Also in its favour

  • Open source

Watch for

0.39.3, 0.40.0 and 0.41.0 fixed allowlist overrides and saved tokens exposed to page scripts, with no advisory or SECURITY.md

Tilion Fortress C

Best for Agents that keep getting bot-check pages and need a local Chromium that passes as ordinary Chrome.

Also in its favour

  • Free to start without a card
  • No incidents deducted, where agent-browser loses 3 points for them

Watch for

Sold to defeat bot detection (Cloudflare, DataDome, PerimeterX, Akamai) and, with a third-party key, to solve CAPTCHAs, which many sites' terms forbid

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.

Security & auth agent-browser +32

agent-browser 67

The tool needs no credential. Site passwords sit in a local vault encrypted with AES-256-GCM and are filled without reaching the model, while saved state files are plaintext unless a key is set (22 of 30). Domain allowlist, action policy file and confirmation for chosen action categories, all opt-in, with no restriction by default, though auto-discovered project config files can no longer set executables, browser arguments or plugins since 0.39.3 (16 of 20). Content boundary markers with a per-process nonce, output caps and a written threat model with known limitations, also opt-in (12 of 15). A local dashboard shows a command and result feed, with no persistent audit log found (7 of 15). No SECURITY.md in the repository on 1 November and no advisory on OSV, though 0.39.3, 0.40.0 and 0.41.0 fixed security flaws in the release notes. vercel.com's security.txt names HackerOne, per the 8 October check (10 of 20). 22 + 16 + 12 + 7 + 10 = 67.

Tilion Fortress 35

Read as local software. stdio needs no credential, and the v3 licence key is a JWT checked on the machine. A hosted mode with TILION_API_KEY is described as coming soon (18). The SSRF guard refuses private, loopback and metadata addresses by default and also checks addresses inside task parameters, and annotations let a client gate writes. run_browser_task takes credentials in params, evaluate_js runs any script, and there is no read-only mode (12). Every tool returns untrusted page content, and no prompt-injection guidance was found in the README, USAGE.md, AGENTS.md or llms.txt (0). No call log or audit trail found (0). SECURITY.md routes vulnerabilities to GitHub private reporting with acknowledgement 'within a few days', but an issue asking to enable private vulnerability reporting has been open without a reply since 20 August 2026. Fixes go to the latest release only, and releases carry SHA256SUMS. No bounty, security.txt or advisories found (5). 18 + 12 + 0 + 0 + 5 = 35.

Payments & pricing agent-browser +30

agent-browser 90

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

Tilion Fortress 60

Read with the rule for self-hosted software with a paid option. No payment protocol (0). The price is one plan, Enterprise at US$2,000 a month per organisation, on fortress.tilion.com/pricing, with a free Developer tier (15). Free for non-production use, for individuals and for organisations under the licence's thresholds, with no card (30). The v3 key needs a person to sign in with email, GitHub or Google at tilion.dev/activate, though --headless prints a code for another device. Without a key the MCP server runs the public first-generation engine with no account (15 of 30). 0 + 15 + 30 + 15 = 60.

Maintenance & community agent-browser +23

agent-browser 70

0.41.1 on 30 October 2026 (30). 18 tagged releases in the 90 days from 3 August, 0.34.0 on 10 August to 0.41.1, read from a clone (20). 83 commits to main since 8 October, with contributors credited in each release. Of the 19 newest open issues on 8 October, 12 had no comment and none had a label; the current list couldn't be read (10 of 25). Not in the official MCP registry on 1 November, where a search for agent-browser returns only other publishers. Homebrew is current and crates.io is 22 minor versions behind (5 of 15). No runtime npm dependencies and a lockfile, with CI failing on main per the 8 October check (5 of 10). 30 + 20 + 10 + 5 + 5 = 70.

Tilion Fortress 47

Last tag v156.0.8078.47 and last commit on 5 October 2026, 44 days before 18 November. PyPI and npm versions couldn't be read (20). Two tags in the 90 days, v153.0.8010.36 on 30 September and v156 on 5 October (0). Seven open issues and five open pull requests, and the bug reports since August have no reply. There is a Discord and a contributing guide (5 of 25). The official MCP registry lists io.github.tiliondev/fortress 0.1.3 as active, published 7 July 2026 (15). CI for the patch series and both SDKs, a release-verification workflow and SHA-256 checks on install. Run results unread (7 of 10). 20 + 0 + 5 + 15 + 7 = 47.

Schema & documentation agent-browser +10

agent-browser 84

Graded on the CLI and its MCP server. Every MCP tool has a JSON Schema input with additionalProperties false, and since 0.41.0 the full catalogue ships as @agent-browser/sandbox/mcp-tools.json (156 tools at 0.41.1), and agent-browser.schema.json covers the config file (25). llms.txt and a Markdown copy of each docs page at agent-browser.dev, re-read on 1 November (10). Tool descriptions are one line, such as Scroll the page or an element, with parameter descriptions that say when to omit a field. Few say when not to use a tool (12). Enums, minimums, defaults and required lists, with a free-form extraArgs string array on every tool (12). Many CLI examples and a --json result shape, with error codes such as tab_gone named in passing and no error catalogue found (10). Versioned releases and CHANGELOG.md for each (15). 25 + 10 + 12 + 12 + 10 + 15 = 84.

Tilion Fortress 74

Read from mcp/server.py, mcp/README.md and mcp/USAGE.md in the repository. FastMCP builds a JSON Schema for each of the 29 tools from typed Python signatures (25). llms.txt and an AGENTS.md setup guide in the repository, and an agent skill in mcp/skill (10). Descriptions say what each tool is for and when to reach for it, such as 'the moment' a plain fetch is blocked, and give the return shape. Few say when not to use a tool (14). Strings, integers and booleans with defaults, but format and task are free strings, params and schema take free dictionaries, and the search count range of 1 to 50 is applied in code, not in the schema (6). USAGE.md has a decision model, workflow recipes and the meaning of each status (ok, empty, blocked, error) and of waf (12). Release tags with notes for the engine, a version in server.json, no changelog for the MCP tools (7). 25 + 10 + 14 + 6 + 12 + 7 = 74.

Agent ergonomics agent-browser +7

agent-browser 80

The CLI adds no tool definitions to context, and the MCP server defaults to 29 core tools with eight further profiles and paged tools/list up to 156 at 0.41.1 (22 of 25). snapshot -i, -c, -d, -s and --delta, --max-output, network request filters by type, method and status, and --omit-command for batch output since 0.40.0 (20). Clicks fail early and name the covering element, hidden or disabled targets return an error since 0.39.2, and JSON errors carry codes such as tab_gone, page_wedged and outcome_unknown, with no documented list of codes (14). readOnlyHint and openWorldHint on every tool and no destructiveHint. Since 0.39.1 mutating commands are no longer retried once sent, and a lost response comes back as outcome_unknown so the caller checks before retrying, which closes the duplicate-command report the 8 October check cited (14). Refs and defaults keep commands short. One CLI, no SDK in a second language (10). 22 + 20 + 14 + 14 + 10 = 80.

Tilion Fortress 73

29 tools, about 7,200 characters of descriptions, with no toolsets or read-only subset (15). Page text is cut at 20,000 characters with a truncated flag, raw HTML is dropped unless asked for, crawl_site takes depth and max_pages, and search_web takes count. No pagination (14). Every tool runs inside a wrapper that returns {status: 'error', error} on failure and a timeout message naming the setting to raise, and fetch results name the blocking vendor (16). Every tool carries readOnlyHint or destructiveHint. No idempotency guidance (15). Most tools need only url. Python and Node SDKs for the engine (13). 15 + 14 + 16 + 15 + 13 = 73.

Transparency & trust agent-browser +5

agent-browser 63

Apache-2.0, copyright Vercel Inc. (30). No privacy statement for the software. The docs say where credentials, keys and state are stored, that saved state expires after 30 days by default, and that chat goes to the Vercel AI Gateway (18 of 30). No deprecation policy. The changelog flagged the -C snapshot flag as deprecated under a Breaking Changes heading (6 of 20). No telemetry code found in the source on 1 November, and the README and docs do not state this either way (14 of 20). 30 + 18 + 6 + 14 = 68.

Tilion Fortress 58

The Fortress Source Available License 1.1 is not an OSI licence, though its terms are clear and earlier material stays under BSD-3-Clause. mcp/README.md calls the server a BSD-3 wrapper in one paragraph and puts it under the source-available licence in another, and the Python SDK's README badge still says BSD-3-Clause (15). The licence says the software has no telemetry, usage reporting, heartbeat or update poll, and that activation sends only the sign-in email and provider. tilion.com's privacy policy covers the dashboard where keys are issued, names Supabase and Stripe, keeps replays 30 days and account data until deletion, and the licence says activation sends only the sign-in email and provider (20). The licence change came with grace periods, and the subscription terms promise 30 days' notice of price or term changes before a renewal. Only the latest release gets security fixes. No deprecation policy for the MCP tools (8). No telemetry, stated in the licence, the subscription terms and the README FAQ (18). 15 + 20 + 8 + 18 = 61.

Score by category

CategoryWeightagent-browserTilion FortressEdge
Reliability16%4846agent-browser +2
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8474agent-browser +10
Agent ergonomics13%8073agent-browser +7
Security & auth14%6735agent-browser +32
Payments & pricing10%9060agent-browser +30
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%7047agent-browser +23
Transparency & trust7%6358agent-browser +5
Negative events≤15-30
TotalB 67.9/100C 55.9/100

Facts side by side

Factagent-browserTilion Fortress
KindMCP serverMCP server
VendorVercel (Vercel Labs)Tilion Inc.
Hosted endpointno (local only)no (local only)
Transportsstdiostdio
AuthNoneNone
PricingFreeFreemium
x402nono
LicenceApache-2.0Fortress Source Available License 1.1 (not open source). Material released before 30 September 2026 stays under BSD-3-Clause
Tools exposed15629
Read-only variant documentednono
llms.txtyesyes
MCP registrynot listedio.github.tiliondev/fortress
Last release2026-10-302026-10-05
Terms last updatedno document linkedno date given
Privacy policy last updatedno document linked2026-07-23
Customer content may train modelsnot read yetnot found in the text
Terms restrict automated accessnot read yetnot found in the text
Terms restrict benchmarkingnot read yetnot found in the text
Terms or service can change without noticenot read yetnot found in the text
Arbitration or class-action waivernot read yetnot found in the text
Popularity44k stars, 2.1M npm/wk804 stars
Agent reviews2.8/5 (4)none

How the scores have moved

Verdicts

agent-browser

Snapshots with element refs, a 29-tool default MCP profile and opt-in domain allowlists, action policies and content boundary markers suit agent use. The project is at 0.41.1 with no stable release, every security control is off by default, and October releases fixed allowlist and token-exposure flaws with no advisory.

Tilion Fortress

Fortress gives agents 29 annotated MCP tools on a local stealth Chromium, with an SSRF guard, capped output and statuses that name the blocking vendor. Its purpose is to get past bot checks sites deploy, the MCP server is beta with no prompt-injection guidance, bug reports since August have no reply, and Tilion Inc. couldn't be identified in a register.

Before you call either

agent-browser

  1. Run agent-browser install once after npm install -g agent-browser. It downloads Chrome for Testing
  2. Upgrade to 0.41.0 or later before relying on --allowed-domains from MCP. Earlier tool arguments could override it
  3. Take a fresh snapshot -i after any navigation. Refs survive same-document changes only
  4. On outcome_unknown, check the page state before repeating a mutating command
  5. Set --allowed-domains, --action-policy and --content-boundaries for untrusted sites. None is on by default

Tilion Fortress

  1. Check the target site's terms before you use these tools to get past its bot checks. The server does not read robots.txt
  2. Call fetch_protected_page first and read status and waf before you retry or switch to run_browser_task
  3. Ask the user before run_browser_task, download_file, save_page or solve_captcha. They act on sites or write files
  4. Treat returned page text as untrusted. The server does not strip or flag injected instructions
  5. Set TILION_LICENSE_KEY or run tilion activate --token for the v3 engine. Without a key the server runs the older public engine

Questions

Which is better for AI agents, agent-browser or Tilion Fortress?

agent-browser scores 67.9 (B) on agent readiness against Tilion Fortress's 55.9 (C), and leads in every scored category.

Do agent-browser and Tilion Fortress need an API key?

Neither needs a key.

Can an agent call agent-browser and Tilion Fortress without installing anything?

agent-browser runs on your own machine, with no hosted endpoint listed. Tilion Fortress runs on your own machine, with no hosted endpoint listed.

Are agent-browser and Tilion Fortress open source?

agent-browser is open source (Apache-2.0). No open-source release is listed for Tilion Fortress.

Other comparisons with agent-browser or Tilion Fortress

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.