Every tool we ship.

$ POST /mcp tools/call db_delete_all
policy: tool not allow-listed
{"error":{"code":-32010,"message":"policy denied"}}
session alive · mcp.tool_denied logged

MCP Proxy

An MCP gateway that runs every JSON-RPC message through an ordered check chain: policy, rug-pull detection, guardrails, redaction, audit. Open source, embeddable in your own agent host.

Kind

Go binary + library

For

MCP gateway

Group

Preventing

the default model
16.7 hours back every week
yearly savings $51,200
38% of one developer's yearly cost

Implementation Savings Calculator

Puts a yearly price on the access work your team does by hand: credential requests, service desk tickets, manual queries. Defaults from a real rollout, every assumption editable.

Kind

Web tool

For

Business case

Group

Business case

paste: delete from users
rule: DELETE without WHERE
blocked · safe query allowed
$ warden test rules/*.yaml

Warden

Build a Hoop guardrail regex from a query you want blocked, test it live against your own examples, and export it ready to enforce. The CLI proves rules in CI.

Kind

Web tool + CLI

For

Guardrails

Group

Preventing

$ blueprint scan --org
RDS · Aurora · DynamoDB · ElastiCache · Redshift
read-only census of every account you can reach
computed locally · zero telemetry

Blueprint

A read-only census of every managed database your AWS credentials can reach: RDS, Aurora, DynamoDB, ElastiCache, Redshift and more. Computed locally, zero telemetry.

Kind

CLI

For

Database discovery

Group

Tooling

$ /plugin install hoop@hooplabs
fence guardrails active
masking · risk reports · token savings
user-level install · no sudo

Hoop Plugin for Claude

The whole toolbelt in one install: Fence guardrails, AI session risk analysis, Julius token savings, and Alcatraz PII masking, wired into every Claude Code session.

Kind

Plugin

For

Claude Code plugin

Group

Tooling

uses: hoophq/alcatraz-action@v1
findings: credit_card · cpf · email
sticky comment posted · values masked
check fails until removed or allow-listed

Alcatraz Action

Detects PII in PR diffs, log files, and issue text before it ships: 45 entity types, checksum validated. Runs entirely in the runner, no external API.

Kind

GitHub Action

For

CI/CD security

Group

Preventing

$ claude "clean up the old records"
> DELETE FROM users WHERE last_login < ...
fence: blocked · destructive query on prod
everything else runs untouched

Fence

Blocks the catastrophic command before your agent runs it, and stays silent for everything else. Semantic guardrails for Claude Code with no denylist to evade.

Kind

Claude Code hook

For

Guardrails

Group

Preventing

DATABASE_URL=
postgres://cloak:a8f2c9e1@localhost:5433/app
real credential swapped on egress
prod-db:5432 · zero secrets in context

Credential Cloak

Hands your agent a fake credential and swaps in the real one on egress, loaded from your OS keychain. The real secret never enters context or logs.

Kind

Local proxy

For

Agent security

Group

Preventing

$ julius run -- go test ./...
command output compressed before it enters context
failures kept verbatim
works with any coding agent

Julius

Cuts token spend where it actually burns: dev-command output entering agent context and API traffic from your scripts. Typically 60 to 90% savings on supported commands.

Kind

CLI + proxy

For

Token efficiency

Group

Tooling

import "github.com/hoophq/alcatraz"
matches := alcatraz.Analyze(text)
email · ssn · credit_card, checksum validated
in-process · no network call

Alcatraz

A dependency-free Go library that detects PII in-process: 45 entity types across 12 countries, checksum validated. No service to deploy, no network call.

Kind

Go library

For

PII detection

Group

Preventing

$ hooprs
security risks · PII exposure · unauthorized access
raw credentials pasted into context
report written · nothing leaves your machine

AI Risk Summary

One command scans your AI session logs locally, surfacing security risks, PII exposure, and unauthorized access before they become incidents.

Kind

CLI

For

Session audit

Group

Preventing

labor · retained revenue · time to value
payback month
90-day · 1-year · 5-year ROI
partial inputs work

AI ROI Calculator

Models labor savings, retained revenue, and time-to-value acceleration into payback plus 90-day, 1-year, and 5-year ROI. Partial inputs work.

Kind

Web tool

For

Business case

Group

Business case