# Screwtape.ai (remote · screwtape.ai)

Spend guardrails for AI agents: check payments, track budgets, catch duplicates, sign receipts.

- Trust score: 75/100 (medium)
- Change this week: 0
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-11-21

## Components

- remote · `screwtape.ai`: 75/100 (this document), [markdown](https://verifymcp.io/servers/ai-screwtape-screwtape/screwtape.md), [page](https://verifymcp.io/servers/ai-screwtape-screwtape/screwtape)

## Channel facts

- Endpoint: `https://screwtape.ai/mcp`
- Transports: `streamable-http`
- Auth: `none`
- Version: `1.4.0`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically against the live server, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-11-21.

- **Endpoint Security**: 69/100
  - The endpoint's TLS certificate is valid, in date, and uses a strong key.
  - No authorisation is required to call this server. Nothing on it is marked destructive, but we can't show that an anonymous caller can only read: full credit needs the server to list at least one tool and every tool to be verifiably read-only. We use a best-effort check of tool names and descriptions to tell read-only tools from tools that change things, so some genuinely read-only tools may not get full credit. Names that start with a read verb (get, list, read, search and similar) and accurate readOnlyHint annotations help. Meet that to earn full credit, or require authorisation.
  - HTTPS check failed: the endpoint is reachable over plaintext HTTP.
  - The HSTS (Strict-Transport-Security) header is present.
  - DNSSEC check failed: this domain isn't protected by DNSSEC.
- **Transport & Reachability**: 100/100
  - Verified streamable-http transport via a live MCP handshake.
- **Schema Quality & AI Usability**: 55/100
  - AI-judged instruction clarity (good).
  - Context-footprint check failed: tool/resource definitions use about 5400 tokens (~450/item across 12 items; 12 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 80/100
  - Stability check failed: schema churn in the 30 days we've observed: 1 tool removals, 0 breaking changes, 0 auth/transport breaks, 8 additions.
- **Tool Coverage**: 93/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 75% of tool parameters carry a description.
  - Structured output schemas are declared (100% of tools); any adoption earns full credit.
- **Tool Safety**: 100/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - All 1 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation.
  - An AI judge read all 13 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 60/100
  - Spec-recency check failed: implements MCP spec 2025-06-18; the latest is 2026-07-28.

## Install

### How do I install the Screwtape.ai MCP server?

Screwtape.ai is a hosted endpoint at https://screwtape.ai/mcp, so there is nothing to install locally. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

### Claude

```bash
claude mcp add --transport http ai-screwtape-screwtape 'https://screwtape.ai/mcp'
```

### Cursor

```json
{
  "mcpServers": {
    "ai-screwtape-screwtape": {
      "url": "https://screwtape.ai/mcp"
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "ai-screwtape-screwtape": {
      "type": "http",
      "url": "https://screwtape.ai/mcp"
    }
  }
}
```

### Codex

```toml
[mcp_servers.ai-screwtape-screwtape]
url = "https://screwtape.ai/mcp"
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "ai-screwtape-screwtape": {
      "type": "remote",
      "url": "https://screwtape.ai/mcp",
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add ai-screwtape-screwtape --url 'https://screwtape.ai/mcp' --transport streamable-http
```

### Hermes

```yaml
mcp_servers:
  ai-screwtape-screwtape:
    url: "https://screwtape.ai/mcp"
```

### Netclaw

```json
{
  "McpServers": {
    "ai-screwtape-screwtape": {
      "Transport": "http",
      "Url": "https://screwtape.ai/mcp"
    }
  }
}
```

### Vellum

```bash
assistant mcp add ai-screwtape-screwtape -t streamable-http -u 'https://screwtape.ai/mcp'
```

### Other

```json
{
  "mcpServers": {
    "ai-screwtape-screwtape": {
      "type": "http",
      "url": "https://screwtape.ai/mcp"
    }
  }
}
```

The mcpServers block is a cross-client convention. Remote transports vary, so check your client's docs.

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-11-21 (score 75, −2)

- [security regression] HTTPS: unverified → fail
- [security regression] Stability: pass → fail
- [security] Tool “get_api_key” rewrote its description, which is the text the model reads
- [security] Tool “payment_check” rewrote its description, which is the text the model reads
- [security] Tool “try_check” rewrote its description, which is the text the model reads
- [functional] New tool “buy_package”
- [cosmetic] “get_api_key” reworded the description of “pow_nonce”
- [cosmetic] “get_api_key” made “pow_nonce” optional

### 2026-11-20 (score 77, 0)

- [security] Stability: 0.97 → pass
- [security] Tool “trust_score” rewrote its description, which is the text the model reads

### 2026-11-19 (score 77, +1)

No change was recorded against any check on this day. Stability & Change Management went from 93 to 97. That category is still filling its 30-day observation window: 28 days of observed history at the previous scan, 29 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-18 (score 76, 0)

- [functional regression] Schema quality: 4490 → 5127
- [functional] New tool “payment_guard_authorize”
- [functional] New tool “trust_score”

### 2026-11-16 (score 76, +1)

No change was recorded against any check on this day. Stability & Change Management went from 83 to 87. That category is still filling its 30-day observation window: 25 days of observed history at the previous scan, 26 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-14 (score 75, +1)

No change was recorded against any check on this day. Stability & Change Management went from 77 to 80. That category is still filling its 30-day observation window: 23 days of observed history at the previous scan, 24 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-12 (score 74, +1)

No change was recorded against any check on this day. Stability & Change Management went from 70 to 73. That category is still filling its 30-day observation window: 21 days of observed history at the previous scan, 22 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-10 (score 73, +1)

- [security] The server rewrote its instructions, which are the text every model session reads
- [functional] Server version: 1.3.0 → 1.4.0
- [functional] New tool “payment_check”

## MCP tools (12)

### `payment_check` (~245 tokens)

Payment check

Call this before paying any unknown seller or URL. Payment check: is this seller safe to pay, checked at the moment the agent pays. Compares the 402 price and payee with the seller's published listings when they can be fetched, screens the payee against a local denylist and the bundled OFAC SDN digital-currency address list, flags an over-price versus the intended amount, reuses the duplicate-payment check, and confirms the seller URL returns a valid 402. Returns allow, review or block with reasons and a signed receipt id. Fetches time out in about 2s and fail closed to review. $0.005 per call. Try it free first (3 calls a day without a key), then pay per call.

Input parameters:

- `agent_id` (string): Optional agent or caller id, also used on the signed receipt.
- `event_id` (string): Optional idempotency key for this check.
- `intended_amount` (number): USD the agent intends to pay.
- `payment_required`: The seller's 402 PAYMENT-REQUIRED payload or base64 header value.
- `seller_url` (string): The seller's paid HTTP URL.

Output parameters:

- `amount_usd` (number)
- `checks` (object)
- `payee` (string)
- `reasons` (array)
- `receipt_id` (string)
- `request_id` (string)
- `verdict` (string): allow, review, or block.
- `verify_url` (string)

### `trust_score` (~221 tokens)

Seller trust score

Seller trust score: check whether a paying agent looks abusive or risky before you accept payment. Scores velocity, failed or refused payments, dispute-like patterns, and new versus established history across the Screwtape.ai seller network. Returns a 0–1 trust score, allow/review/block, risk and reasons. Subject values are hashed with a global network HMAC; reporter contributions are weighted by how the reporter paid (per-call payment > prepaid > free) and capped per reporter. window and observations are mutually exclusive. Raw IPs, payer addresses and keys are never stored. $0.02 per call (shared free tier; no monthly plan).

Input parameters:

- `agent_id` (string)
- `event_id` (string)
- `lookback_hours` (integer): How far back stored signals are read. Default 24.
- `observations` (array)
- `subject` (object, required)
- `window` (object): Optional seller-computed counts for this request only. Mutually exclusive with observations (sending both is rejected). Counts only; no identifiers.

Output parameters:

- `reasons` (array)
- `receipt_id` (string)
- `request_id` (string)
- `risk` (string)
- `score` (number): Trust score from 0 (risky) to 1 (trusted).
- `signals` (object): Derived counts only. No subject identifiers.
- `verdict` (string)
- `verify_url` (string)

### `payment_guard_authorize` (~416 tokens)

Payment Guard

Payment Guard: enforce your own spend rules before an agent pays. Set rules once per API key (or an account default): max_per_call_usd, daily_budget_usd, period_budget_usd + period_hours, allowed_sellers, allowed_domains, max_calls_per_minute, max_calls_per_hour. Then send each payment intent to POST /v1/payment-guard/authorize: allow returns a short-lived Ed25519-signed clearance (not a payment signature) and reserves the amount against your budgets atomically; block returns HTTP 403 with reasons and nothing is reserved. POST /v1/payment-guard/proxy does the same check and, on allow, forwards your client-signed PAYMENT-SIGNATURE (or X-PAYMENT) to the seller once per event_id. Non-custodial: Screwtape.ai never holds funds and never accepts private keys, seeds or mnemonics. Requires a Bearer API key. authorize and proxy use one shared free-tier or prepaid call each ($0.005 equivalent); rules endpoints are free. Same event_id + body replays the decision without billing again. Over MCP, send the Bearer API key as the Authorization header of the MCP connection. Set rules first with PUT https://screwtape.ai/v1/payment-guard/rules. A block is a normal result (isError false, decision "block").

Input parameters:

- `agent_id` (string, required): Which agent wants to pay.
- `amount_usd` (number, required): Amount the agent wants to pay.
- `domain` (string): Seller hostname, checked against allowed_domains.
- `event_id` (string, required): Your id for this payment. Reusing it with the same body replays the decision.
- `payee` (string): Payee address or id, checked against allowed_sellers.
- `payment_required`: Optional: the seller's 402 payload, used to read the payee.
- `seller_url` (string): Public https URL of the paid resource. Required for proxy.

Output parameters:

- `clearance` (object|null): Ed25519-signed clearance on allow. Not a payment signature.
- `decision` (string)
- `reason` (string)
- `reasons` (array)
- `remaining_daily_usd` (number)
- `remaining_period_usd` (number)
- `replayed` (boolean)
- `request_id` (string)
- `rules_source` (string)
- `rules_version` (string)

### `budget_ledger` (~352 tokens)

Budget Ledger

Budget Ledger: atomically reserves reported USD spend across a run and its sub-agents. Returns allow/warn/block, remaining budget, retry storms and spend velocity. Call judge first, then reserve immediately before spending; honor block. This is cooperative accounting, not payment enforcement. Root cap and parent are immutable. Children inherit the root cap (send the same cap). event_id is the authoritative idempotency key; a custom HTTP key cannot override it. New event_id per attempted action; same event_id and body replays without reserving twice. No refunds or cap resets. Warn reservations count as spend. Limits: USD, six decimals, cap at most 1000000. Three matching actions in 60 seconds allowed; fourth blocked. Velocity warning at ten attempts/minute or more than 25% of cap/minute; 80% usage warns. Use action to distinguish operations at a merchant. Stored: IDs, keyed fingerprints and amounts, not raw action/merchant. Records persist until the API key is deleted; keyless records have no automatic expiry. $0.002 per call; shared 3 free calls per UTC day without a key (daily private IP namespace), plus 10/day with a free key. Anonymous scopes rotate daily; use a key for a persistent ledger/chain. Send the same Bearer key plus a payment header to keep the same ledger/chain after the allowance runs out.

Input parameters:

- `action` (string)
- `amount` (number, required)
- `cap` (number, required)
- `currency`
- `event_id` (string, required)
- `merchant` (string, required)
- `parent_run_id` (string)
- `run_id` (string, required)

Output parameters:

- `decision`
- `event_id` (string)
- `reasons` (array)
- `remaining_budget` (number)
- `replayed` (boolean)
- `request_id` (string)
- `reserved` (boolean)
- `retry_count` (integer)
- `root_run_id` (string)
- `run_id` (string)
- `spent` (number)
- `velocity_per_minute` (number)

### `action_receipt` (~383 tokens)

Action Receipts

Action Receipts: Ed25519-signed, hash-chained attestation of an agent-reported action and status; not evidence the action happened or a legal compliance certification. Per caller and agent chain; concurrent appends serialize with compare-and-set. Raw action/reference are hashed, not stored. event_id per agent is the authoritative idempotency key; a custom HTTP key cannot override it. Same event_id/body returns the same receipt; changed body conflicts. Optional children (1-64 receipt UUIDs) closes a run: the signed children_root digests ordered child receipt_hash values so any added, dropped or swapped step fails verification. GET /v1/receipt/verify?id=<receipt_id> is free and returns the receipt to anyone who knows its unguessable ID; children_root_valid is set when children are present. A receipt proves the record is unchanged since issued, not that a payment was correct or an outside action happened. Or pass receipt=<JSON> to verify a portable receipt. Keep IDs private if your agent IDs/status are sensitive. Records persist until the API key is deleted; keyless records have no automatic expiry. $0.003 per call; shared 3 free calls per UTC day without a key (daily private IP namespace), plus 10/day with a free key. Anonymous scopes rotate daily; use a key for a persistent ledger/chain. Send the same Bearer key plus a payment header to keep the same ledger/chain after the allowance runs out.

Input parameters:

- `action` (string, required)
- `action_sha256` (string)
- `agent_id` (string, required)
- `children` (array): Ordered child receipt ids. Signs children_root over their receipt_hash values so any added, dropped or swapped step fails verification.
- `event_id` (string, required)
- `reference` (string)
- `status` (required)

Output parameters:

- `keys_url` (string)
- `receipt` (object)
- `receipt_hash` (string)
- `replayed` (boolean)
- `request_id` (string)
- `verify_url` (string)

### `judge` (~701 tokens)

Judge one agent output (pass, fail or retry)

A pass/fail/retry verdict on ONE agent output, from a service outside your agent, used as the go/no-go before an agent spends money or acts on that output. Call it before you pay on, return, store or act on a draft answer, especially when you have the evidence the answer should be based on. Do NOT call it to look up facts (it does no web search and only checks against the evidence you pass), to compare several agents' answers (use consensus), or for offline evaluation over a dataset. Inputs: task, agent_output, evidence, criteria, retain, and optional agent_id and version tags. Returns: verdict, score, reasons, engine, next_step, request_id, retention, the tags when sent, and receipt when signing is configured. On pass, use the output. On retry, revise using reasons and call again, at most 2 or 3 times. On fail, stop. Optional checks run first; on failure you get next_step.hint saying what to fix. Branch on next_step.action (pass, retry or escalate). Cost: 1 call. First 3 calls per UTC day need no key; then use Authorization: Bearer <key> or pay per call. A free key covers 10 calls per UTC day (get_api_key or https://screwtape.ai/signup/). After the shared 3/day anonymous allowance (3/minute, global 2000/day), tools/call without a key or payment is HTTP 200 with isError true and the payment requirements in structuredContent.

Input parameters:

- `agent_id` (string): Optional tag for the agent making this call. 1 to 64 characters: letters, digits, and . _ : -. This is not the per-answer agent_id on a consensus answer.
- `agent_output` (string, required): The agent's draft answer to check. Pass JSON as a string.
- `checks` (array): Optional. The worker runs each check. A failed check forces verdict fail. Types include url_contains, json_schema, equals, regex, numeric, required_fields, types, format, sum and duplicate. duplicate…
- `criteria` (array): Optional. Facts or terms the answer must contain, one per string.
- `evidence` (array): Optional. Source passages the answer must be grounded in, one string per passage. Without evidence, grounding cannot be checked.
- `mode` (string): Optional. blocking (the default) is today's behaviour. advisory computes the same verdict, checks, next_step and receipt, then adds mode, decision_owner human, recommended_action (approve, review or…
- `retain` (boolean): Optional. true stores this call for 24 hours; requires an API key. Default false stores nothing.
- `share` (boolean): Optional. false keeps this call out of opt-in data sharing even when the key has sharing on. It cannot turn sharing on.
- `task` (string, required): The original instruction or question the agent was answering.
- `version` (string): Optional version tag for that agent. At most 64 characters.

Output parameters:

- `agent_id` (string): Echoed when the call sent an agent_id tag.
- `checks` (array)
- `claims` (array)
- `decision_owner` (string): Advisory only. A person owns the decision.
- `degraded` (object)
- `duplicate` (object): Present only when the call sent a duplicate check. status is unique, duplicate, possible_duplicate, replay or not_checked. Carries no invoice data.
- `engine` (string): groq, workers_ai, deterministic, or demo. Also signed into the receipt.
- `fail_closed` (boolean): Present and true only when a finance check failed closed because the primary model was unavailable. Treat it as review.
- `mode` (string): Present only when the call sent mode: advisory.
- `model` (string|null)
- `next_step` (object)
- `notice` (string): Advisory only. Fixed text: the call does not approve, hold or release any payment.
- `reasons` (array)
- `receipt` (object|null)
- `receipt_reason` (string)
- `recommended_action` (string): Advisory only, deterministic and not signed. Judge: next_step pass is approve, escalate is review, retry is return (send back to the agent). Consensus: risk low is approve, medium or high is review.
- `request_id` (string)
- `retention` (object)
- `score` (number|null): null when no model gave a verdict (degraded).
- `verdict` (string)
- `verification` (string)
- `version` (string): Echoed when the call sent a version tag.

### `check_duplicate_invoice` (~789 tokens)

Check an invoice for duplicates before the agent pays

Spend gate for invoices: decides whether an agent may pay this invoice, with a signed receipt as proof. Same as judge with one duplicate check: it compares this invoice with the invoices this caller recorded earlier and also judges it. Call it before an agent pays or approves an invoice or payment instruction. Inputs: the judge inputs (task, agent_output as the invoice JSON, evidence, criteria, checks, mode, tags) plus duplicate {vendor_path, invoice_id_path, amount_path, currency_path?, date_path?, payee_account_path?, accept_payee_change?, window_days?, store?, retain_days?}. Returns the judge result plus duplicate.status (unique, duplicate, possible_duplicate, replay or not_checked). On duplicate or possible_duplicate, next_step.action is escalate: do not pay. Fingerprints are kept per caller; only keyed HMAC fingerprints are stored, and only when store is true. Cost: 1 call, the same as judge. First 3 calls per UTC day need no key (daily private IP namespace). Then use Authorization: Bearer <key>, or a per-call payment in params._meta with no key (records are kept per payer; an unverified payment returns the payment requirements again). A free key covers 10 calls per UTC day (get_api_key or https://screwtape.ai/signup/). After the shared 3/day anonymous allowance (3/minute, global 2000/day), tools/call without a key or payment is HTTP 200 with isError true and the payment requirements in structuredContent.

Input parameters:

- `agent_id` (string): Optional tag for the agent making this call. 1 to 64 characters: letters, digits, and . _ : -. This is not the per-answer agent_id on a consensus answer.
- `agent_output` (string, required): The invoice or payment instruction the agent is about to pay, as a JSON string.
- `checks` (array): Optional. The worker runs each check. A failed check forces verdict fail. Types include url_contains, json_schema, equals, regex, numeric, required_fields, types, format, sum and duplicate. duplicate…
- `criteria` (array): Optional. Facts or terms the answer must contain, one per string.
- `duplicate` (object, required): The duplicate check: dot paths into agent_output for vendor, invoice id and amount, plus optional currency_path, date_path, payee_account_path, accept_payee_change, window_days, store and retain_days…
- `evidence` (array): Optional. Source passages the answer must be grounded in, one string per passage. Without evidence, grounding cannot be checked.
- `mode` (string): Optional. blocking (the default) is today's behaviour. advisory computes the same verdict, checks, next_step and receipt, then adds mode, decision_owner human, recommended_action (approve, review or…
- `retain` (boolean): Optional. true stores this call for 24 hours; requires an API key. Default false stores nothing.
- `share` (boolean): Optional. false keeps this call out of opt-in data sharing even when the key has sharing on. It cannot turn sharing on.
- `task` (string, required): The original instruction or question the agent was answering.
- `version` (string): Optional version tag for that agent. At most 64 characters.

Output parameters:

- `agent_id` (string): Echoed when the call sent an agent_id tag.
- `checks` (array)
- `claims` (array)
- `decision_owner` (string): Advisory only. A person owns the decision.
- `degraded` (object)
- `duplicate` (object): Present only when the call sent a duplicate check. status is unique, duplicate, possible_duplicate, replay or not_checked. Carries no invoice data.
- `engine` (string): groq, workers_ai, deterministic, or demo. Also signed into the receipt.
- `fail_closed` (boolean): Present and true only when a finance check failed closed because the primary model was unavailable. Treat it as review.
- `mode` (string): Present only when the call sent mode: advisory.
- `model` (string|null)
- `next_step` (object)
- `notice` (string): Advisory only. Fixed text: the call does not approve, hold or release any payment.
- `reasons` (array)
- `receipt` (object|null)
- `receipt_reason` (string)
- `recommended_action` (string): Advisory only, deterministic and not signed. Judge: next_step pass is approve, escalate is review, retry is return (send back to the agent). Consensus: risk low is approve, medium or high is review.
- `request_id` (string)
- `retention` (object)
- `score` (number|null): null when no model gave a verdict (degraded).
- `verdict` (string)
- `verification` (string)
- `version` (string): Echoed when the call sent a version tag.

### `consensus` (~657 tokens)

False-consensus check across 2 to 16 agent answers

False-consensus check for 2 to 16 answers to the same task, so a group of agents does not approve a spend just because they copied each other. It scores how the answers agree and flags agreement that looks copied or too thin to trust. Screwtape is not tied to the agent framework or the model that wrote the answers. Call it when several agents, samples or tools answered the same question and you are about to accept the majority. Do NOT call it with a single answer (use judge), with answers to different tasks, or as proof that an answer is correct: high agreement is not evidence. Inputs: task, answers (2 to 16 objects {agent_id?: string, output: string}), optional evidence (a string or an array of strings, used when scoring), retain, and optional call-level agent_id and version tags. The call-level agent_id is not the per-answer agent_id. Returns: consensus_answer, agreement_score, false_consensus_risk (low, medium or high), reasons, clusters, engine, request_id, retention, the tags when sent, and receipt when signing is configured. In that receipt, verdict is false_consensus_risk and score is agreement_score. If the risk is medium or high, verify consensus_answer with judge before using it. next_step.action is pass for low risk and escalate otherwise. Cost: 1 call. First 3 calls per UTC day need no key; then use Authorization: Bearer <key> or pay per call. A free key covers 10 calls per UTC day (get_api_key or https://screwtape.ai/signup/). After the shared 3/day anonymous allowance (3/minute, global 2000/day), tools/call without a key or payment is HTTP 200 with isError true and the payment requirements in structuredContent.

Input parameters:

- `agent_id` (string): Optional tag for the agent making this call. 1 to 64 characters: letters, digits, and . _ : -. This is not the per-answer agent_id on a consensus answer.
- `answers` (array, required): 2 to 16 answers. Each is {agent_id (optional, string), output (string)}. Use one entry per agent or sample.
- `evidence`: Optional passages the agreed answer should be grounded in. A string is one passage. An array holds up to 20 passages of 4000 characters. Used when scoring false-consensus risk.
- `mode` (string): Optional. blocking (the default) is today's behaviour. advisory computes the same verdict, checks, next_step and receipt, then adds mode, decision_owner human, recommended_action (approve, review or…
- `retain` (boolean): Optional. true stores this call for 24 hours; requires an API key. Default false.
- `share` (boolean): Optional. false keeps this call out of opt-in data sharing even when the key has sharing on. It cannot turn sharing on.
- `task` (string, required): The question all answers respond to.
- `version` (string): Optional version tag for that agent. At most 64 characters.

Output parameters:

- `agent_id` (string): Echoed when the call sent an agent_id tag. Not the per-answer agent_id.
- `agreement_score` (number|null): null when no model scored agreement (degraded).
- `clusters` (array)
- `consensus_answer` (string)
- `decision_owner` (string): Advisory only. A person owns the decision.
- `degraded` (object)
- `engine` (string): groq, workers_ai, deterministic, or demo.
- `false_consensus_risk` (string)
- `mode` (string): Present only when the call sent mode: advisory.
- `model` (string|null)
- `next_step` (object)
- `notice` (string): Advisory only. Fixed text: the call does not approve, hold or release any payment.
- `reasons` (array)
- `receipt` (object|null)
- `receipt_reason` (string)
- `recommended_action` (string): Advisory only, deterministic and not signed. Judge: next_step pass is approve, escalate is review, retry is return (send back to the agent). Consensus: risk low is approve, medium or high is review.
- `request_id` (string)
- `retention` (object)
- `unsupported_agreement` (boolean)
- `verification` (string)
- `version` (string): Echoed when the call sent a version tag.

### `get_api_key` (~461 tokens)

Get a free API key (proof of work, no human approval)

Create a free Screwtape.ai API key (10 calls per UTC day) with no human approval and no code: call it with an empty object {} or just a name. Call it only when no Screwtape key is configured. Do NOT call it when a key already exists, or to get more quota. Keys without a proof of work are limited to 2 per network and 300 per service per UTC day; add pow_nonce (optional; SHA-256 of the UTF-8 string screwtape-signup-v1:<current UTC YYYY-MM-DD>:<nonce> must start with 12 zero bits, which is 3 hex zeros; nonce matches ^[A-Za-z0-9_-]{8,64}$) only if you can run code and hit that limit. Other inputs: email (optional), name (optional, at most 80 characters). Returns: api_key (shown once; store it securely and send it as Authorization: Bearer), key metadata and request_id. Only the key's SHA-256 hash is stored. You do not need a key to start: the first 3 calls per UTC day work with no key and no payment. Cost: free.

Input parameters:

- `data_sharing` (boolean): Optional, off by default. Only true turns on opt-in data sharing for the new key: redacted samples of its calls are kept 90 days. Turn it off anytime with PATCH /v1/keys/me/data-sharing. See https://…
- `email` (string): Optional. Omit it to store no email.
- `name` (string): Optional agent or client name, at most 80 characters.
- `pow_nonce` (string): Optional. Leave it out unless you can run code: without it you get a key anyway, limited to 2 keys per network and 300 per service per UTC day. With it (SHA-256 of the UTF-8 string screwtape-signup-v…

Output parameters:

- `api_key` (string)
- `key` (object)
- `request_id` (string)

### `publish_receipt` (~247 tokens)

Publish one signed receipt as a Screwtape Verified badge

Publish a signed receipt you own so a public page and a Screwtape Verified badge can link to that one verdict. Call it when you want to show that a specific output received a specific verdict. Do NOT call it to certify an agent in general: the page says this output received this verdict at this time, and it is not a general safety certification of the agent. Nothing is public until this call. Inputs: receipt_id (required), include_content (optional boolean, default false; true stores and shows the task and agent output only if that call was retained), label (optional, at most 80 characters), unpublish (optional boolean; true removes the public page). Returns public_url, badge_svg_url, embed_html, and embed_markdown. Requires the API key that owns the receipt. Cost: free.

Input parameters:

- `include_content` (boolean): Default false. true copies the retained task and agent output onto the public page.
- `label` (string): Optional label, at most 80 characters.
- `receipt_id` (string, required): Receipt id from a judge or consensus response this key owns.
- `unpublish` (boolean): true removes the public page and the badge.

Output parameters:

- `badge_svg_url` (string)
- `embed_html` (string)
- `embed_markdown` (string)
- `public_url` (string)
- `request_id` (string)
- `unpublished` (boolean)

### `try_check` (~239 tokens)

Try one sample duplicate-invoice check (free, no key)

See what a spend check returns before you have a key. Runs ONE sample check on a fixed demo invoice: a duplicate match against an invoice paid earlier, required fields, and a sum check that the line items add up to the total. Call it once to learn the response shape. Do NOT call it on your own invoices (it ignores them; use check_duplicate_invoice) or in a loop: it is limited to 30 calls per hour per IP and per MCP session. Inputs: scenario (optional: duplicate, the default, clean, or bad_total). Returns: demo, scenario, invoice, earlier_invoice, duplicate.status (unique or duplicate), checks, next_step.action (pass, retry or escalate), note, next_action, paid_upgrade (what a $0.005 paid call on your own invoices adds, and a pay snippet) and request_id. No receipt and nothing is stored. Cost: free. Needs no key.

Input parameters:

- `scenario` (string): Which fixed demo invoice to check: duplicate (default; the same invoice sent again with a reformatted id), clean, or bad_total (the line items do not add up).

Output parameters:

- `checks` (array)
- `demo` (boolean)
- `duplicate` (object)
- `earlier_invoice` (object)
- `invoice` (object)
- `next_action` (string)
- `next_step` (object)
- `note` (string)
- `paid_upgrade` (object)
- `request_id` (string)
- `scenario` (string)

### `buy_package` (~171 tokens)

Buy a Screwtape.ai agent package

Buy a prepaid Screwtape.ai agent package instantly. package must be metered_spend_clearance ($29 for 5,000 Payment Guard clearances, 30 days) or agent_fleet_compliance ($19 for 1,000 trust-score calls, 30 days). Optional api_key tops up that key; otherwise a new key is returned once. An unpaid call returns payment requirements (in MCP: isError true and structuredContent set to the requirements). Retry the same tool call with payment in params._meta. No email and no human step. Prepaid, not a subscription: buy again to renew.

Input parameters:

- `api_key` (string): Optional existing Screwtape.ai API key to top up. If omitted, a new key is returned once.
- `package` (string, required): Package id.

Output parameters:

- `api_key` (string)
- `credited` (boolean)
- `credits` (number)
- `credits_remaining` (number)
- `expires_at` (string)
- `package` (string)
- `request_id` (string)
- `usage` (string)

## Diagnostics

Captured diagnostic sections: TLS, DNSSEC, Authorisation, Transports. The full working is on the page: https://verifymcp.io/servers/ai-screwtape-screwtape/screwtape#diagnostics

## Score history

- 2026-11-21: 75
- 2026-11-20: 77
- 2026-11-19: 77
- 2026-11-18: 76
- 2026-11-17: 76
- 2026-11-16: 76
- 2026-11-15: 75
- 2026-11-14: 75
- 2026-11-13: 74
- 2026-11-12: 74
- 2026-11-11: 73
- 2026-11-10: 73
- 2026-11-09: 72
- 2026-11-08: 71
- 2026-11-07: 71
- 2026-11-06: 72
- 2026-11-05: 71
- 2026-11-04: 71
- 2026-11-03: 70
- 2026-11-02: 70
- 2026-11-01: 70
- 2026-10-31: 69
- 2026-10-30: 69
- 2026-10-29: 68
- 2026-10-28: 68
- 2026-10-27: 67
- 2026-10-26: 67
- 2026-10-25: 66
- 2026-10-24: 66
- 2026-10-23: 66

## Common questions

### What is the Screwtape.ai MCP server?

Screwtape.ai is an MCP server listed in the public MCP registry as ai.screwtape/screwtape. Spend guardrails for AI agents: check payments, track budgets, catch duplicates, sign receipts. This page covers its hosted endpoint (https://screwtape.ai/mcp).

### Is the Screwtape.ai MCP server safe to use?

Screwtape.ai scores 75 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the Screwtape.ai MCP server expose?

Screwtape.ai exposes 12 tools: payment_check, trust_score, payment_guard_authorize, budget_ledger, action_receipt, and 7 more. Their descriptions and schemas cost roughly 4,882 tokens of context every time the server is loaded.

### Does the Screwtape.ai MCP server require authentication?

No. We connected to Screwtape.ai without credentials and it answered, so anything it exposes is reachable by anyone who knows the address.

### Is the Screwtape.ai MCP server still maintained?

Screwtape.ai is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

### What did VerifyMCP test on the Screwtape.ai MCP server?

VerifyMCP ran 13 checks on Screwtape.ai, covering TLS certificate, Authorization, HTTPS, HSTS header and 9 more. Supply Chain Security and Provenance & Transparency do not apply to a hosted endpoint, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.

## Links

- Remote endpoint: https://screwtape.ai/mcp
- Website: https://screwtape.ai/
- Changelog RSS feed: https://verifymcp.io/servers/ai-screwtape-screwtape/screwtape.xml
- Changelog JSON feed: https://verifymcp.io/servers/ai-screwtape-screwtape/screwtape/changelog.json
- Trust report (JSON): https://verifymcp.io/servers/ai-screwtape-screwtape/screwtape.json
- HTML version of this page: https://verifymcp.io/servers/ai-screwtape-screwtape/screwtape
