Home › Search: guardrails

Search: guardrails

Search MCP servers and agent skills by name, description, category or topic — 19 results.

Agent Skill Active

mattpocock/skills

17 dev workflow skills: PRD writing, TDD, codebase architecture, git guardrails, issue triage, refactoring plans, and more

198.9k Shell Updated yesterday Score 85
Agent Skill Active

alinaqi/claude-bootstrap

Opinionated project initialization with security-first guardrails, spec-driven atomic todos, LLM testing patterns, and CLI tool orchestration (gh, vercel, supabase)

705 Python Updated 18d ago Score 78
MCP Server Active

juspay/neurolink

Making enterprise AI infrastructure universally accessible. Edge-first platform unifying 12 providers and 100+ models with multi-agent orchestration, HITL workflows, guardrails middleware, and context summarization.

117 TypeScript Updated today Score 71
MCP Server Official Active

get-tmonier/argot

Guardrail that checks agent-written code against a repo's own git-history-learned patterns — flags foreign dependencies, reinvented functions, misplaced code, broken layering, and gamed tests. Exposes `voice_context` (the repo's familiar imports/idioms for a file, so agents write in-voice before generating), `check` (score a generated hunk against the repo's patterns), and `explain`. 100% local, n

12 Rust Updated today Score 71
MCP Server Active

taniwhaai/arai

Policy enforcement for AI coding agents derived from existing instruction files (CLAUDE.md, .cursorrules, .windsurfrules, .github/copilot-instructions.md) — no separate YAML to maintain. Rules with prohibitive predicates (`never`, `forbids`, `must_not`) emit `permissionDecision: deny` to block tool calls in Claude Code; advisory rules inject context. PostToolUse is correlated with PreToolUse to pr

7 Rust Updated 5d ago Score 59
MCP Server Active

AperionAI/shield

Local guardrail proxy for AI coding agents. Wraps any MCP server (stdio or Streamable HTTP) and blocks destructive tool calls — DROP TABLE, rm -rf, force-push — before they execute. MCP supply-chain protection: TOFU tool-catalog pinning against rug pulls, plus tool-description and tool-result scanning for tool poisoning and prompt injection. 51 starter rules, approval gates, audit logging. Single

6 Rust Updated 4d ago Score 58
MCP Server Active

s-b-e-n-s-o-n/portkey-admin-mcp

MCP server for the Portkey AI Gateway Admin API — 150 tools for prompts, configs, analytics, keys, guardrails, integrations, and more.

5 TypeScript Updated 4d ago Score 58
MCP Server Active

kiro0x/five-mcp

LLM character consistency engine — generates structured JSON constraints from 4 multiple-choice questions about an AI's psychology. Drop the JSON into any LLM's system prompt to prevent persona drift; reduces inference cost from retries. 160,000 personality patterns; works with any LLM.

5 Python Updated 20d ago Score 58
MCP Server Active

1luvc0d3/metabase-mcp

MCP server connecting Claude to Metabase with 28 tools for natural language data analysis, dashboard management, SQL queries, and automated insights. Features SQL guardrails, rate limiting, and audit logging.

4 TypeScript Updated 20d ago Score 57
MCP Server Active

bmdhodl/agent47

Runtime guardrails and incident read access for coding agents. Query AgentGuard traces, alerts, usage, costs, and budget health.

4 Python Updated 3d ago Score 57
MCP Server Active

zyx77550/sparda

Injects a live, reversible MCP server into a running Express / FastAPI / Next.js app — reads safe by default, writes gated behind human confirmation. The same engine also proves deploys and PRs (`apocalypse` / `review`).

4 JavaScript Updated 2d ago Score 57
MCP Server Active

Jott2121/agent-gate

Lets an AI agent gate its own work before claiming "done": a fail-closed ship checklist (`verify_gate` counts a check satisfied only if explicitly proven), required independent refute-first review, and an append-only sha256-chained receipts ledger that makes any edit to past receipts detectable. `pip install mcp-agent-gate`.

3 Python Updated 3d ago Score 56
MCP Server Active

ThinkneoAI/mcp-server

ThinkNEO Control Plane — Enterprise AI governance MCP server with runtime guardrails, observability, AI FinOps, and agent lifecycle control.

3 Python Updated 26d ago Score 56
MCP Server Active

loicfontaine-max/qorami-sdk

Check an email before an AI agent sends it: returns send / ask-a-human / block, with machine reason codes and prompt-injection detection.

2 Python Updated 29d ago Score 55
MCP Server Active

nonameuserd/paybond-kit

Tenant-bound MCP server for agent spend authorization, policy guardrails, evidence, escrow settlement, and portable receipts. Sandbox bootstrap, spend intents, and completion validation for Cursor and Claude Code. TypeScript: `npx -p @paybond/kit paybond login` then `npx -y -p @paybond/kit paybond-mcp-server`. Python: `pip install paybond-kit` then `paybond-kit-login` and `paybond-mcp-server`. [Do

1 TypeScript Updated 3d ago Score 53
MCP Server Active

jamjet-labs/jamjet-policy

MCP stdio interceptor (`@jamjet/mcp-shim`) that applies one YAML policy file (block / require_approval / audit / budget cap) to `tools/call` requests before they reach the real MCP server. The same policy also runs in Claude Code PreToolUse hooks (`@jamjet/claude-code-hook`), OpenAI Agents SDK guardrails (`@jamjet/openai-guardrail`), and JamJet's Python/TS SDKs — `jamjet audit show` tails every de

2 TypeScript Updated 19d ago Score 50
MCP Server Maintained

narekmalk/safedb-mcp

Secure MCP server for safe, read-only DB access by AI agents, with SQL guardrails, table allowlists, PII masking, and audit logs.

7 TypeScript Updated 2mo ago Score 47
MCP Server Active

lumayapartners/memini

Local-first project memory and guardrails for coding agents. Records failed attempts, fragile files, and decisions per repo; warns the agent before it repeats a recorded mistake. `npx -y memini mcp`

0 TypeScript Updated 10d ago Score 45
MCP Server Active

bluetieroperations-create/blackwall-mcp

Pre-action risk gate for AI agents. One `forecast` tool the agent calls before any irreversible action (send money, run SQL, delete data); returns a risk score (0–100), reversibility class, named red flags from 28 failure modes, and a gate: proceed / confirm / human-required.

0 JavaScript Updated 19d ago Score 40