One subscription · Browser · REST · MCP

Independent infrastructure for work AI agents should not improvise.

ScrubMyText gives people and agents deterministic tools for content safety, consequential actions, and evidence-backed service selection.

Three product families, one platform

Choose the outcome you need. Every agent operation remains precisely scoped and independently callable.

AI agent guides

Start with the problem—not a product name.

Get a direct answer, implementation boundary, working starting point, and honest build-versus-buy comparison for common agent failures.

Browse all agent guides →

Built for the full agent lifecycle

Start with a focused tool and keep the same platform as the work becomes more consequential.

Before

Prepare and decide

Inspect content, validate outputs, check policies and terms, and compare contextual service evidence.

During

Constrain and coordinate

Lock duplicate actions, enforce budgets and deadlines, obtain independent authority, and wait for outside events.

After

Prove and recover

Record receipts, reconcile outcomes, compensate failed actions, revoke stale context, and preserve correction state.

For builders

One API key

Every paid plan unlocks the same 49-operation portfolio with hard usage limits and no surprise overages.

See plans →
For agents

Answers agents can discover

Direct problem answers, honest comparisons, OpenAPI, MCP metadata, llms.txt, quickstarts, and structured catalogs stay synchronized.

Agent answers and comparisons →
For people

Focused browser workflows

Use individual content tools and public trust cards without navigating the API catalog or writing code.

Browse browser tools →

What “independent infrastructure” means

The model proposes work. A separate, deterministic service decides whether that work may proceed and records what happened.

Example 1

A retry should not become a second payment

An agent can repeat a tool call after a timeout even when the first call succeeded. LockMyAction derives the same key from the intended action and gives only one caller the right to proceed. It does not move money or decide whether a purchase is wise; it only makes the retry boundary explicit.

Example 2

Untrusted text should not silently become instructions

PrepareMyContext normalizes hidden characters, reports suspicious instruction patterns, detects likely secrets, and can redact common sensitive values. The report is evidence for a caller's policy—not a promise that arbitrary content is safe.

Example 3

A service rating needs context

TrustMyChoice separates observed task evidence from a universal popularity score. A service can be reliable for one operation and poorly evidenced for another. Cards expose sample size, recency, uncertainty, and correction state so an agent can apply its own threshold.

Deterministic where it matters

Locks, counters, deadlines, digests, and schema checks have explicit inputs and outputs. They do not depend on a model producing the same judgment twice.

Small contracts, not orchestration

ScrubMyText is not an agent framework, browser, model router, or automation platform. It supplies narrow controls that can sit inside any of those systems through REST or MCP.

Evidence, not guarantees

Detection tools report defined signals and documented limits. They complement application authorization, secure storage, malware scanning, and human judgment rather than replacing them.

Company, support & policies

Clear information for people using free tools or subscribing to the platform.