MCP REGISTRY FOR AGENTS DOCS PRICING CONTROL YOUR AGENTS

// MCP SERVER RECORD

Looking Glass

Unverified

NPMlooking-glass-mcp

FRisk gradecritical blast radius

Catalogue entry updated 5 days ago

Auth postureNot probedno remote endpoint probed yet

Tools718 Destructive · 29 Execute · 2 Write · 22 Read · 10 Other

Worst categoryDestructivegrade tracks the peak, not the average

Capability mix

Destructive 8Execute 29Write 2Read 22Other 10

What it can reach

Ingests untrusted inputTouches secretsReaches networkRuns codeTouches filesChanges permissionsSends external commsPersistsDeletes dataMoves money

Lethal trifectaReads attacker-controllable content, holds secrets, and can run code + reach the network — a full exfiltration/RCE path if unconstrained.

Tools 71

browser_clean_slateDestructiveHigh browser_clear_historyDestructiveHigh browser_clear_mocksDestructiveMedium browser_clear_storageDestructiveHigh browser_closeDestructiveMedium

View all 71 tools

session_endDestructiveHigh test_chaos_clearDestructiveHigh vault_deleteDestructiveHigh browser_clickExecuteHigh browser_dragExecuteHigh browser_evaluateExecuteCritical browser_exploreExecuteHigh browser_fill_and_submitExecuteHigh browser_goExecuteHigh browser_go_backExecuteMedium browser_go_forwardExecuteMedium browser_hoverExecuteMedium browser_loginExecuteHigh browser_mock_routeExecuteHigh browser_navigateExecuteHigh browser_performance_auditExecuteMedium browser_press_keyExecuteHigh browser_recoverExecuteMedium browser_resolve_elementExecuteMedium browser_scrollExecuteMedium browser_select_optionExecuteMedium browser_smart_clickExecuteHigh browser_tab_newExecuteMedium browser_tab_selectExecuteMedium browser_typeExecuteHigh browser_wait_forExecuteMedium browser_wait_until_stableExecuteHigh browser_workflowExecuteHigh session_startExecuteHigh test_chaosExecuteHigh test_fill_formExecuteMedium test_scenario_runExecuteHigh browser_set_cookieWriteMedium browser_set_localstorageWriteMedium browser_action_historyReadLow browser_analyze_pageReadLow browser_console_messagesReadLow browser_diff_stateReadMedium browser_error_reportReadLow browser_extractReadMedium browser_get_cookiesReadLow browser_get_localstorageReadLow browser_health_checkReadLow browser_list_mocksReadLow browser_network_requestsReadLow browser_screenshotReadLow browser_snapshotReadLow browser_snapshot_stateReadLow browser_tab_listReadMedium session_listReadLow session_replayReadMedium test_accessibility_auditReadMedium test_scenario_statusReadLow vault_listReadLow visual_baselineReadLow visual_compareReadLow browser_diagnoseOtherLow browser_suggest_actionsOtherLow session_export_playwrightOtherMedium test_assertOtherLow test_auth_flowOtherMedium test_generate_assertionsOtherLow test_stop_watchOtherLow test_watch_eventsOtherLow vault_injectOtherHigh vault_loginOtherMedium

These definitions cost 5,080 tokens of context →

Change history

No change history on record for this server.

Recommended policy

browser_clean_slatedeny

browser_clickrequire approval

browser_set_cookierate limit

Read-only toolsallow

Full policy breakdown with enforcement rules →

This record as markdown: /tools/io-github-sahib-sawhney-wh-looking-glass-mcp.md — append .md to any tool or server page.

// RUNNING THIS SERVER

DIRECT INSTALL npx -y looking-glass-mcp

Installed this way, nothing enforces the recommended policy above — calls run exactly as the agent makes them.

// LOOK UP ANOTHER SERVER

Every MCP server has a record like this.

Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.

Teams ship this data inside their own products. See what a licence covers →