Agent Enhancer Utilities

Free reliability tools for AI agents to plan and recover guarded workflows, coordinate work, test failure paths, create safe fixtures, and debug integrations.

Watch video demo Copy a starter prompt

Quick start

Connect, copy, and test

Pick your agent and copy one starter prompt. It contains the correct no-auth endpoint and introduces the service as an orchestration sidecar. A direct ChatGPT install link will be added here after marketplace approval. See compatibility and open-source install options →

ChatGPT

Guide →

Paste once to connect the helper and start a new task.

Connect prompt

Use Agent Enhancer Utilities as a reliability sidecar for this chat. If it is not connected, tell me to add this no-auth MCP server: https://liberated.site/mcp/chatgpt?source=website-home-chatgpt. Use the smallest relevant tool only when it improves orchestration, coordination, or recovery. Never send secrets or personal data. Briefly explain the sidecar idea, then ask what task I want to do.

Claude

Guide →

Paste once to connect the helper and start a new task.

Connect prompt

Use Agent Enhancer Utilities as a reliability sidecar for this chat. If it is not connected, tell me to add this no-auth MCP server: https://liberated.site/mcp/claude?source=website-home-claude. Use the smallest relevant tool only when it improves orchestration, coordination, or recovery. Never send secrets or personal data. Briefly explain the sidecar idea, then ask what task I want to do.

Codex or any MCP agent

Guide →

For Codex and clients that support remote MCP.

Connect prompt

Use Agent Enhancer Utilities as a reliability sidecar for this chat. If it is not connected, tell me to add this no-auth MCP server: https://liberated.site/mcp?source=website-home-mcp. Use the smallest relevant tool only when it improves orchestration, coordination, or recovery. Never send secrets or personal data. Briefly explain the sidecar idea, then ask what task I want to do.

Available utilities

Browse all tools

coordination

PennyLock

Acquire an opaque distributed lock with a short TTL.

View tool →

coordination

Global Seen Stamp

Atomically record whether an opaque content hash has been seen.

View tool →

coordination

Exactly-Once Baton

Issue and atomically consume a one-use capability token.

View tool →

coordination

Negative Cache Ticket

Record or check a typed, expiring failure state.

View tool →

ephemeral

Status-Code Forge

Create a short-lived URL with controlled HTTP behavior.

View tool →

fixtures

Safe Synthetic Fixture Vault

Generate deterministic test records using reserved non-production values.

View tool →

compatibility

Error-Code Cemetery

Map a covered integration-error fingerprint to a tested diagnosis.

View tool →

x402-intelligence

402 Error Rosetta Stone

Translate a covered x402 failure fingerprint into a correction.

View tool →

x402-intelligence

Worked-Once Recipe Vault

Return a covered, empirically verified integration recipe.

View tool →

coordination

Swarm Semaphore

Bound concurrent agent work with an opaque distributed semaphore.

View tool →

coordination

Barrier Bell

Release a bounded agent phase after enough unique participants arrive.

View tool →

coordination

Freshness Lease

Coordinate one renewable freshness owner without exposing identity.

View tool →

ephemeral

Failure Sequence Forge

Create a short-lived URL that atomically advances through a retry sequence.

View tool →

coordination

Swarm Rate Gate

Share a bounded, retry-safe token budget across an agent swarm.

View tool →

compatibility

MCP Tool Contract Linter

Lint one bounded MCP tool contract for review and interoperability risks.

View tool →

x402-intelligence

x402 Requirement Drift Diff

Compare two normalized x402 requirements before retrying a payment.

View tool →

ephemeral

Webhook Attempt Meter

Create a bounded webhook endpoint and retrieve attempt metadata once.

View tool →

compatibility

MCP Capability Handshake Diff

Explain effective MCP capabilities and negotiation mismatches.

View tool →

x402-intelligence

x402 Facilitator Compatibility Diff

Match one x402 requirement to supplied facilitator capabilities.

View tool →

compatibility

MCP Elicitation Safety Linter

Flag sensitive MCP elicitation fields that should use URL mode.

View tool →

coordination

x402 Quote Fingerprint Guard

Bind one normalized x402 quote to an opaque retry identity.

View tool →

compatibility

MCP Schema Edge-Case Atlas

Return bounded MCP JSON Schema portability test vectors.

View tool →

workflow-reliability

Workflow Guard Planner

Plan a bounded reliability sidecar for an external agent workflow.

View tool →

workflow-reliability

Opaque Workflow Checkpoint

Fence and recover an opaque external workflow without claiming external proof.

View tool →

Agent workflows

Guard repeated external plugin workflows

Agents get a deterministic guard plan, generation-fenced recovery state, explicit uncertainty handling, and an honest guarantee without sending external records to the sidecar.

View workflow →

Coordinate parallel agents without duplicate work

Workers coordinate through opaque identifiers and TTL-bounded state without accounts, a shared database, or exposing task contents.

View workflow →

Test HTTP failure paths safely

Agents can reproduce client behavior using deterministic status responses and clearly synthetic data with published expiry and content limits.

View workflow →

Debug x402 integrations with evidence

Agents get a free coverage decision, evidence-linked diagnosis, and an explicit next check before changing payment code.

View workflow →

Review MCP tool contracts before submission

Agents can find concrete contract and negotiation defects before a connector scan while keeping credentials, field values, and production payloads out of the review.

View workflow →

Guard x402 retries against changed terms

Agents can stop a retry when commercial or resource terms drift and distinguish document compatibility from live facilitator behavior.

View workflow →

Measure webhook retries without storing bodies

Agents can verify bounded webhook attempt and payload-stability behavior without retaining source IPs, request bodies, raw user agents, or arbitrary headers.

View workflow →

Inspect. Validate. Execute safely.

Find the smallest tool for the job.
Check its schema, limits, and side effects.
Use the documented retry identity and recovery path.