"Loop" matching MCP servers:

Enables detection of autonomous agent infinite tool loops, planning deadlocks, and liveloss cycles via trajectory state entropy analysis. Also provides safety guardrails for prompt injection, destructive commands, budget overruns, and data leakage.

Navigator for BPC Agentic Procurement, providing tools for submitting RFQs, checking status, requesting terms, and escalating to a human.

Enables autonomous agents and MCP-compatible clients to enforce policy-approved actions using risk tiers, reversible checkpoints, and token approval bounds.

Enables agents to query real-time labelled pump.fun intelligence—launch risk cards, creator reputation, rug/graduation flags, wallet profiles, and market regime—with paid endpoints settled per request via x402 USDC on Solana and free sample/regime tools.

Runtime governance and budget guardrails for Claude Code, Cursor, and autonomous AI agents. Enforces per-session spend caps, verifier safety gates, and runaway loop prevention.

Gives local and cloud LLMs live web grounding as four MCP tools: web_search, fetch, deep_search (a token-capped, cited evidence pack sized to a small context window), and research (an agentic search-and-read loop). Flat monthly pricing, no query content stored.

An MCP server that enables human-in-the-loop elicitation, letting AI agents ask users questions via tools like elicit_confirm and elicit_form, with diagnostics to verify host elicitation support.

Human-in-the-loop approval inbox for AI agents: an agent proposes an action (send email, post comment, run a command), a human approves, rejects, or edits it from a web, mobile, or Slack/Discord/Telegram inbox, and the agent only runs on approval. Full audit trail, self-hostable (MIT).

Enables AI coding agents to request human-in-the-loop approvals, choices, refinements, and notifications via MCP, with fail-closed semantics and no remote approval surface.

Lets agents query their own usage ledger captured by the Agentic Ledger proxy: sessions with full call traces, costs, loop runs, and stuck-loop flags. Tools include list_sessions, get_session, search, list_runs, get_run_status, and explain.

MCP server that gives an AI agent a throwaway test identity — a real disposable email address and a real UK phone number — so it can sign up for something it's testing and read back the verification email/SMS itself, without a human in the loop.

Enables creating scheduled Gantt charts with critical-path analysis from any MCP client using plain-language task descriptions, with no account or API key required.

Enables AI agents to search merchant catalogs, verify product prices and stock, and create purchases with budget limits and human-in-the-loop approval in Russian specialty stores.

Human-in-the-loop automation framework. Log them in, watch them run.

Architectural sensor for Python codebases. Scores structural health (modularity, acyclicity, depth, equality), detects import cycles, enforces YAML layer rules, and runs a snapshot/diff loop so AI-assisted edits do not silently regress structure.

WhatsApp notifications and human-in-the-loop for AI agents. Text "join" to get an API key, then send messages, ask questions with tap-to-answer buttons, and read replies - no Meta account, no templates, no dashboard.

Enables MCP clients to evaluate spend gates and manage human-in-the-loop approvals for CHP Profile B.

Product management over a shared product spine — link customer feedback and insights to features, tasks, sprints and releases, then read closed-loop analytics (customer 360, funnels, retention, NRR) on the same records. Local stdio server via npx @aioproductoscom/mcp; also a 38-tool hosted remote (OAuth 2.1).

An MCP server for in-loop design review of web previews. It enables agents to submit a preview URL, receive structured findings with suggested fixes, and recheck after applying changes, while never editing code itself.

Provides full control of a macOS system, enabling AI agents to execute shell commands, run AppleScript, manage files and processes, and perform human-like GUI interactions such as clicking, dragging, scrolling, and typing, with a screenshot-based perception loop.