TAP

tech.humanPublished by human.techOfficial publisherv0.1.0Updated Oct 24, 2026

Credential isolation for AI agents: placeholder secrets, policy checks, optional human approval.

VerifiedStreamable HTTPWeb executableDeveloper ToolsSecurity & Monitoring

Overview

AI-generated overview

TAP is a hosted remote MCP endpoint that isolates AI agents' credentials using placeholder secrets, policy checks and optional human approval.

What you can do
  • Route agent requests through connectors with credential isolation, keeping real secrets out of the agent.
  • Substitute placeholder secrets for real credentials and validate their position in requests.
  • Enforce approval policies, with optional human approval flows and Telegram or Matrix approval bots.
  • Audit request routing and connector behavior, and debug failed requests.
When to use it
Use it when an assistant needs to call external services with real credentials but you want those secrets kept out of the agent, with policy checks or human sign-off before sensitive requests go through.
Requirements
A remote streamable HTTP endpoint at mcp.tap.human.tech. OAuth authentication is required, plus a TAP agent API key supplied in the X-TAP-Key header, created at tap.human.tech. The hosted version is recommended; self-hosting requires Postgres and your own security hardening.
Before you install
The server handles real credentials and signing keys on your behalf, so trust in the host matters. Self-hosting puts host hardening, key isolation and correct policy-engine operation on you and is meant for security-experienced teams. Requests may be gated by approval flows, and the hosted service is a third party that holds your credentials in its enclave.

Installation

In SourceWeft

  1. Open TAP in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/tech.human%2Ftap/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "tap": {
      "type": "http",
      "url": "https://mcp.tap.human.tech/mcp"
    }
  }
}

README

Tool Authorization Protocol (TAP)

Credential isolation, approval gating, and connector routing for AI agents.

This repository contains the code that is most useful for:

  • auditing request routing and connector behavior
  • debugging failed requests
  • understanding approval flows
  • improving connector-side request shaping
  • contributing fixes to the core TAP experience

[!WARNING] Self-hosting means you own the security of your credentials and signing keys. TAP keeps secrets out of your agents, but running it yourself puts the host hardening, key isolation, and correct policy-engine operation on you. It's a path for teams that are well versed in security. For everyone else the hosted version is strongly recommended: credentials sit in a hardware enclave we can't read into, with no ops to run. Start free at tap.human.tech.

Start Here

  • crates/tap-proxy/src/routing.rs — how TAP resolves connector target shapes
  • crates/tap-proxy/src/placeholder.rs — credential substitution and position validation
  • crates/tap-proxy/src/policy.rs — approval policy enforcement
  • docs/ — full documentation including self-hosting guide

Included

  • core proxy and storage crates
  • Telegram and Matrix approval bots
  • remote MCP server (tap-mcp)
  • CLI
  • docs (self-hosting, API reference, credential setup)

Not Included

  • enclave deployment glue (CCE policy generation, release-policy automation, ARM templates, env config)
  • production workflows and secret bootstrapping
  • managed hosting operations glue
  • the hosted dashboard UI source (a placeholder is shipped so the proxy compiles)

The enclave key-management source is included (key_provider_enclave.rs, kms_azure.rs, skr.rs) — it's the custody model documented at docs.tap.human.tech/security, and each hosted release's enclave measurement is published in measurements/. Hosted deployment and operational infrastructure are maintained separately from this repository.

Security

See SECURITY.md to report a vulnerability.

License

Apache-2.0: free to use, read, modify, and self-host. This repo is the open-source TAP runtime (tap-core, tap-proxy, tap-bot, tap-cli, tap-mcp). The hosted dashboard and managed-service deployment glue are proprietary and live in a separate private repo.

Contributing

See CONTRIBUTING.md.

Testing

bash
# Needs Postgres (default postgres://tap:tap@localhost:5434/tap, override with# POSTGRES_DATABASE_URL). Isolated suites parallelize; env-mutating unit tests stay serial.cargo test -p tap-corecargo test -p tap-proxy --test integration --test e2ecargo test -p tap-proxy -p tap-bot -p tap-cli --lib --bins -- --test-threads=1

Source: README.md at commit 850d3c9

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

1
  1. v0.1.0LatestSep 16, 2026

Score

85% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

Other Human integrations

UGC, the Human Ad Network

africa.ugcOfficial publisher

Remote MCP endpoint for UGC, a "Human Ad Network", letting an assistant work with ad campaigns, clipping and Clip Studio from chat.

Streamable HTTPWeb executableAuth
Business & CommerceCredential requiredOct 24, 2026

Woodstitch

com.woodstitch

Runs paid, blinded human listening tests on text-to-speech audio and returns preference or human-vs-AI scores with confidence intervals.

STDIODesktop only
Data & AnalyticsNot validated yetNov 7, 2026

J-Dawgs AI Tool Store & Human Hands — Catalog MCP

Published by GitHub account joshuajdawg0350

Read-only catalog server for the J-Dawgs AI Tool Store and Human Hands for AIs, letting an assistant search tools, prices and bundles and learn how agents…

STDIODesktop only
Productivity & Workflow7 toolsNov 4, 2026

ActualHumanTake

Published by GitHub account magiautonomous

Submits a product for an honest, human-written review and lets an assistant check status or read the finished feedback using a secret receipt token.

Streamable HTTPWeb executable
Productivity & Workflow3 toolsNov 1, 2026

HUMAN API

Published by GitHub account human-api-labs

HUMAN API lets an AI assistant find voluntary people and request real-world tasks it cannot do digitally, then check whether a human accepted or declined.

Streamable HTTPWeb executable
Location & Lifestyle3 toolsOct 30, 2026