autopilot-mcp (TylerFlar/autopilot-mcp) is an MCP server listed on the M8ven Trust Index. It scores 52 out of 100, grade D. It declares 39 tools. No publisher has claimed this listing.
A browser-automation MCP server providing persistent browser profiles per domain, Bitwarden credential injection without exposing passwords, and playbook recording/replay for repeatable tasks.
Caution. Specific findings reduced this grade. They are listed on the page. Grades reflect the full trust pyramid: code, verification depth, and reputation. New projects cap at C until adoption is earned.
How we verified
Verified is a snapshot. Live keeps it current, and builds your track record.
⚡ Connect GitHub → continuous verification on every pushwhy connect →Who stands behind it
TylerFlar
Source: Glama
Claim it to get a verified publisher badge, a free copy of our full audit findings, and direct contact for any high-priority issues we find. Or connect your repo for our deepest verification, Live Monitored: read-only, revoke anytime. What we access →
Install from
The grade above is for the source repository. Registries can serve a different version, so we mark the ones we were not able to read.
These names and descriptions are the publisher's own, read from the source code. We print them as written. Our assessment is the findings above, not this list.
spawn_instanceCreate an isolated temporary browser instance for parallel work.
list_instancesList live spawned browser instances.
close_instanceClose a spawned browser instance and delete its temporary profile.
instance_navigateNavigate a spawned instance to a URL.
instance_screenshotPNG screenshot of the page open in a spawned instance.
instance_get_textVisible text from the current page in a spawned instance.
instance_get_urlCurrent URL for a spawned instance.
instance_run_jsRun JavaScript in a spawned instance and return the stringified result.
instance_clickClick viewport coordinates in a spawned instance.
instance_type_textType text into the focused element in a spawned instance.
instance_attach_fileAttach a local file to an input in a spawned instance.
instance_scrollScroll a spawned instance by roughly one viewport.
instance_fill_loginFill a login form in a spawned instance with a Bitwarden entry.
navigateOpen `url` in a persistent browser profile and return the page text.
screenshotPNG screenshot of the page open in `profile`.
get_textVisible text from the current page — cheaper than a screenshot.
get_urlCurrent URL for the profile's page.
run_jsRun JavaScript in the page and return the stringified result. PREFERRED over click() for form filling and button activation — CSS/DOM selectors are robust to viewport size.
clickClick at viewport coordinates. Use only when you can't target by selector — prefer run_js for robustness.
type_textType text into the currently-focused element. Click or focus first.
attach_fileAttach a local file to a ``<input type="file">`` element on the page.
scrollScroll the page by roughly one viewport.
list_loginsSearch Bitwarden for saved logins. Returns id, name, urls, username for each match — NEVER the password. `query=""` lists everything.
fill_loginFill a login form with a Bitwarden entry. Preferred over typing the password yourself — the password is injected into the DOM and never returns to you.
get_totpCurrent 6-digit TOTP for a Bitwarden item. Bitwarden is the single source of truth for TOTP secrets. Codes are ~30s; call right before you need to paste.
reveal_credentialsESCAPE HATCH — returns plaintext username + password. Only use when fill_login can't target the form. `reason` is mandatory and audited.
create_loginCreate a new Bitwarden Login item. Errors if `name` already exists — use update_login or upsert_login to modify an existing entry.
update_loginPatch fields on an existing Bitwarden Login. Errors if `id_or_url` doesn't resolve to a unique item. Leave fields empty to skip — only the fields you pass get written.
upsert_loginCreate a new login, or update the existing one matching (url, username). The "I just signed up, remember these" path. Errors if more than one existing item matches both url and username.
delete_loginDESTRUCTIVE — sends a Bitwarden Login item to trash. Requires `confirm=True`. Recoverable from Bitwarden's trash until purged.
serve_local_filePublish a local file at an unguessable URL on 127.0.0.1 with CORS so the browser can ``fetch()`` it. Use when the site's upload widget is pure JS (drag-drop, fetch+FormData) and there's NO ``<input type="file">`` you can target with ``attach_file``.
list_served_filesList files currently published by ``serve_local_file`` (token, url, path, size, expires_at). Useful when an earlier upload session left a URL active and you want to find or revoke it.
unserve_local_fileRevoke a URL published by ``serve_local_file`` (frees the token immediately rather than waiting for TTL).
list_playbooksList saved playbooks. ALWAYS check first — if a playbook exists for your task, run it instead of re-discovering navigation.
run_playbookExecute a saved playbook. Returns screenshots + extracted text from observation steps. On failure, returns the error and the failed step index.
playbook_run_getReturn the saved JSON ledger for a playbook run.
playbook_run_listList saved playbook run ledgers, newest first.
save_playbookSave a playbook — a reusable sequence of browser steps. Call after a successful first-time task so future attempts are 1 call.
delete_playbookDelete a saved playbook. Use for unfixably-broken playbooks — otherwise save_playbook with the same name overwrites.
HEADLESStrue Set "false" to show the browser window for debugging.BROWSER_TIMEOUT30000 Per-page navigation/action timeout, in ms.AUTOPILOT_BW_TIMEOUT_SECONDS45 Timeout for a single bw CLI invocation.BITWARDENCLI_APPDATA_DIROverride the Bitwarden CLI data directory (standard bw variable).AUTOPILOT_LOG_LEVELINFO Root log level for all autopilot. loggers.AUTOPILOT_TOOL_TIMEOUT_SECONDS60 Wall-clock cap on a single tool call.AUTOPILOT_PLAYBOOK_TIMEOUT_SECONDS300 Wall-clock cap on a run_playbook call.AUTOPILOT_FILE_SERVER_HOST/ AUTOPILOT_FILE_SERVER_PORT env vars.AUTOPILOT_FILE_SERVER_PORTAUTOPILOT_FILE_SERVER_HOST / env vars.Tool annotations
No tools have read-only/destructive annotations
Add readOnlyHint or destructiveHint annotations to every tool so hosts can warn users before invoking.
All four hints declared on every tool
39/39 tools missing one or more hints — spawn_instance (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint); list_instances (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint); close_instance (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint), +36 more. OpenAI's directory rejects tools where any of the four hints are missing or non-boolean.
For every tool, set all four hints (readOnlyHint, destructiveHint, idempotentHint, openWorldHint) to explicit true/false values that match the handler’s actual behaviour.
Destructive tools are labelled
1 tool perform destructive updates without destructiveHint — spawn_instance deletes at line 303 (shutil.rmtree(dest_dir, ignore_errors=True))
Add destructiveHint:true to any tool whose handler calls .delete(), .upsert(), .update(), unlink, rm, DELETE, DROP, REPLACE INTO, or any operation that overwrites existing data.
Tool inputs are validated
37/39 tool handlers declare input schemas (95%)
Declare an inputSchema with zod/joi/yup on every tool definition.
Tool handlers catch errors
Only 3/39 tool handlers wrap calls in try/catch (8%)
Wrap each tool handler body in try/catch and return a structured error response.
License file
No license file
Add a LICENSE file (MIT, Apache-2.0, etc.).
Tests exist
No test files found
Add tests that exercise each declared tool.
Claim the listing to review these findings one by one and send us a correction where you disagree, straight to the team. Claiming also means we tell you when the grade moves, and reach you first if we find anything urgent.
[](https://m8ven.ai/mcp/tylerflar/autopilot-mcp)?variant=verified from the URL.Vetting this one by hand? Tool Check is an MCP that scores other MCPs. Add it once and ask Claude, ChatGPT, or any MCP client to grade a server, surface CVEs, check the publisher, and suggest safer alternatives — before you install.
https://m8ven.ai/api/mcp/tool-check