Real-time MCP & A2A prompt-injection, jailbreak, and tool-schema poisoning firewall for AI agents.
Scored across 2 tools
The two tools are functionally distinct: one reads wallet/billing state and the other scans text for prompt injections, so an agent is unlikely to confuse them. However, the descriptions are bloated and refer to 'the other tools on this server for execution' and 'use get_agent_wallet_status to check...', implying a broader toolset that does not exist, which adds avoidable confusion.
Both names use a snake_case verb_object pattern (get_agent_wallet_status, scan_prompt_injection), which is readable and fairly consistent. The verbs differ slightly in structure (get vs scan) but this is a minor deviation rather than a real inconsistency.
Only two tools exist, and one of them is a billing/quota-status helper rather than a security capability. For a server branding itself as a prompt-injection, jailbreak, and tool-poisoning firewall, this is far too thin a surface to earn its stated scope.
The server advertises jailbreak and tool-poisoning protection, yet only scan_prompt_injection performs any detection and there are no tools for jailbreak checks, tool-poisoning inspection, or sanitization configuration. The only other tool is ancillary billing status, leaving major advertised capabilities uncovered.
Inspect an agent's account ledger: remaining free trial calls (out of 5), prepaid USD balance, cumulative spend, total API calls, and total LLM tokens saved. WHEN TO USE: Call before a batch of paid tool executions or after completing a Stripe or USDC top-up to confirm your balance. WHEN NOT TO USE: Do not use to process web content, execute marketplace tools, or submit payments; use the other tools on this server for execution. PARAMETER INTERACTIONS: Omit agent_id to inspect the default trial account derived from your client IP, or pass your explicit agent_id or agd_pass_ session key to check a specific funded wallet. Free and read-only: never charges a fee or consumes a free trial call. Example: get_agent_wallet_status(agent_id='example')
| Name | Required | Description | Default |
|---|---|---|---|
| agent_id | No | Optional Agent ID or agd_pass_ session key to inspect. Omit to check the default wallet derived from your client IP. |
| Name | Required | Description |
|---|---|---|
| data | No | Structured tool output payload. |
| status | Yes | Execution status ('ok' or 'payment_required'). |
| roi_proof | No | Token compression and cost-savings verification metrics. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint, idempotentHint, and destructiveHint=false, so the safety profile is covered. The description adds real context beyond that: it never charges a fee or consumes a free trial call, and the default wallet is derived from the caller's client IP. It does not discuss rate limits or caching, but the cost/free-trial guarantee is meaningful behavioral disclosure.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Front-loaded with the ledger contents, then structured WHEN TO USE / WHEN NOT TO USE / PARAMETER INTERACTIONS / Example. The labeled sections make it scannable, though the repetition of the read-only guarantee and the worked example add a little bulk.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
An output schema exists, so return-value detail is optional, yet the description still names the key fields. With a single optional parameter, no nested objects, and annotations covering safety, an agent has everything needed to invoke it correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the baseline is 3. The description goes slightly further by naming the agd_pass_ session-key form and framing the two modes as an interaction (omit for IP-derived trial account vs. pass explicit agent_id/session key for a funded wallet), which aids selection even though the schema states the omit behavior.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb (inspect) and resource (agent's account ledger) and enumerates exactly what is returned: trial calls out of 5, prepaid USD balance, cumulative spend, total API calls, tokens saved. This clearly separates it from the sibling web-content tools (distill_webpage, extract_structured_data, recon_webpage_skeleton), which operate on webpages rather than wallet state.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicit WHEN TO USE (before paid tool batches, after a Stripe or USDC top-up) and WHEN NOT TO USE (not for processing web content, executing marketplace tools, or submitting payments). The exclusion clause routes the agent back to the other server tools for execution, so the alternative is named.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Scans untrusted web pages, emails, MCP responses, or A2A payloads for hidden prompt injections and returns sanitized text. WHEN TO USE: Call scan_prompt_injection when your task requires deterministic prompt-guard analysis for text; use get_agent_wallet_status to check your remaining free trial quota or prepaid balance. WHEN NOT TO USE: Do not use for unrelated domains or binary media. PARAMETER INTERACTIONS: Pass the target text as a non-empty string; omit agent_id and tx_hash during your 5 free trial calls, or supply agent_id (agd_pass_...) / tx_hash for paid execution. Read-only and idempotent; results are cached for 1 hour (<5ms cache hit). $0.005 per call. Pricing: the first 5 calls per agent_id are free and failed calls are never charged. After 5 calls, the tool returns a payment-required quote (not a fatal error) with a $2 Stripe top-up link and Base/Solana USDC addresses; complete payment, then retry with your agd_pass_ key in agent_id or pass tx_hash. Example: scan_prompt_injection(text='example')
| Name | Required | Description | Default |
|---|---|---|---|
| text | Yes | Untrusted text, HTML, or tool output to scan for hidden prompt injection or jailbreak payloads. | |
| tx_hash | No | Optional on-chain Base Mainnet (0x...) or Solana SPL USDC transaction hash. Omit unless settling an individual call directly on-chain instead of using an agd_pass_ key in agent_id. | |
| agent_id | No | Optional agent identifier or prepaid agd_pass_ session key. Omit during your 5 free trial calls to use an ID automatically derived from your client IP, or pass your agd_pass_ key after topping up. |
| Name | Required | Description |
|---|---|---|
| data | No | Structured tool output payload. |
| status | Yes | Execution status ('ok' or 'payment_required'). |
| roi_proof | No | Token compression and cost-savings verification metrics. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations cover the safety profile (readOnly, idempotent, non-destructive, openWorld), and the description goes well beyond them: 1-hour result caching with <5ms hits, $0.005 per call, 5 free calls per agent_id, failed calls never charged, and the non-fatal payment-required quote with top-up paths and retry procedure. It correctly and explicitly confirms read-only/idempotent behavior, consistent with annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Well front-loaded: purpose first, then WHEN TO USE / WHEN NOT TO USE / PARAMETER INTERACTIONS, ending with an example. It is dense and the pricing paragraph is long, but each clause addresses a real failure mode (paying incorrectly or retrying wrongly) rather than padding.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a 3-parameter tool with a formal output schema and full annotations, everything an agent needs is present: input expectations, the monetization gate and how to recover from it, caching behavior, and safety profile. Return-value details are correctly left to the output schema.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so the baseline is 3; the description adds routing value by stating the decision rule for agent_id/tx_hash (omit both during the 5 free calls, supply agd_pass_ or a tx_hash for paid execution). That is mostly mirrored in the schema descriptions, so it is additive but not novel.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The first sentence names a specific verb (scans), the exact input classes (web pages, emails, MCP responses, A2A payloads), the target (hidden prompt injections) and the outcome (returns sanitized text). It is immediately distinguishable from the only sibling, get_agent_wallet_status, which handles quota/balance.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicit WHEN TO USE and WHEN NOT TO USE blocks, plus a named alternative (get_agent_wallet_status) for checking quota. Exclusions are concrete: unrelated domains and binary media. No inference required.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
get_agent_wallet_statusscan_prompt_injectionSecurity firewall for AI agents — scans MCP calls for injection, secrets, and risks.
The WAF for agents. Pattern-based + heuristic firewall scans prompts, RAG documents, tool argume...
AgentGuard — 20-tool AI safety MCP: policy preflight, risk scoring, audit logging, rate limits.
Deterministic runtime safety for AI agents: scan PII, gate tool actions, verify LLM output.
Add one secure layer between your agents and this server.
