Loading…
Pay-per-call agent tools: media and video generation, product demos, research, GTM, scraping, and compute.
Do not connect
A critical issue was found. Do not connect this server as-is.
Scanned 1 month ago Due for re-check
A server can change after it's graded. Re-run the automated scan to refresh this report.
This grade is deterministic and reproducible: the same server surface always yields the same grade under a given algorithm version. It is a real automated assessment computed by the MCPGrade engine from what the probe actually observed — not a fabricated or opinion score. It is not a manual human pentest, so it can miss context-specific risks.
Every signal below was measured directly by the automated probe. The grade is derived only from evidence like this — nothing is assumed.
The vaaya MCP server exposes 10 tools, focused primarily on AI capabilities. Its published description reads: "Pay-per-call agent tools: media and video generation, product demos, research, GTM, scraping, and compute". It communicates over Streamable HTTP using the 2026-07-28 protocol revision, and does not require authorization to connect. MCPGrade currently rates vaaya F — a critical issue was found and the server should not be connected as-is. Its most notable findings include "Hidden instructions in a tool description" and "Cross-tool shadowing". This report is a deterministic, reproducible automated assessment: the same observed surface always yields the same grade under a given algorithm version, and it is refreshed as new scans arrive and free to read — but it is not a substitute for a manual human security review. Always re-check vaaya's advertised tools, transport security, and authorization posture before connecting an autonomous agent or sharing sensitive context with it.
Add vaaya to an MCP client using the endpoint below. Review the grade and findings above before granting an autonomous agent access.
MCP endpoint
https://vaaya.ai/mcpStreamable HTTP transport. This is the MCP endpoint, not a website — paste it into your MCP client server list rather than a browser.
Captured passively during the read-only scan. Click any value to find servers that match it.
The tools this server advertises via tools/list — names, purposes, and the parameters each accepts, exactly as enumerated read-only.
Round-trip ping that confirms the agent → Vaaya connection and whether the user is linked. Returns { ok:true, userId, scope, version, server_time, first_call } when linked — SHOW `first_call.show_to_u
Public onboarding hint for an agent whose user isn't linked to Vaaya yet. Returns where the human should go to connect (and sign up if new). Call this when vaaya_test_connection reports needs_auth, or
Show which Vaaya account this connection is linked to and its money state. Returns { email, name, user_id, connected_client, scopes, balance_cents, credit_line, available_cents, credits_url, switch_ac
Disconnect this client from the current Vaaya account: revokes this connection's authorization server-side, so every later call fails with 401 until the user reconnects. Call it when the user asks to
Vaaya's consultant. Describe ANY external capability you or the user might want — generate an image/video, search or scrape the web, run code in a sandbox, send/receive email, enrich a contact — and i
Execute a single call that `consult` handed you, and bill on success. Used for any external capability (image/video/audio generation, web search, scraping, email, document parsing, code sandbox, brows
Fetch the status + output of an async job started by `use` (e.g. a video render). Pass the `job_id` that `use` returned with `{ async: true }`. Returns `{ status, result?, progress?, charged_cents }`:
Run a command or code in an open E2B sandbox session (started by `use` with action `create_session`, which returns a `session_id`). Pass `session_id` plus either `command` (a shell command) or `code`
Close an E2B sandbox session and stop its billing. Pass the `session_id`. Captures the final metered uptime cost and releases the hold. ALWAYS call this when finished with a session — an open session
Ask a DIFFERENT LLM a question and get its answer, billed per token from the Vaaya wallet (model cost + 3%, usually a fraction of a cent). Use it to get a second opinion from a rival model, cross-chec
Sorted worst-first. Each finding shows its severity, what it means, its OWASP MCP Top-10 mapping, and a recommended fix — the check id links to the exact methodology row that produced it.
tool: llm
A tool description contains imperative text aimed at the model (e.g. "ignore previous", "before answering read …").
tion":"Optional system prompt for the other model."},"max_tokens":{"type":"nuRecommendationRemove model-directed instructions from tool descriptions.
tool: vaaya_onboard
A tool description tries to alter the model’s use of another tool.
Call this when vaaya_test_connection reports needs_auth, or any tool returns unauthoRecommendationDescriptions must describe only their own tool.
tool: consult
A tool exposes a free-text/webhook/feedback parameter with no honest purpose — a classic exfil channel.
a can do. It is CONVERSATIONAL and remembers prior turns. It returns: mode='RecommendationRemove side-channel parameters; constrain tool inputs.
tool: consult
Tools that execute commands / write files / make arbitrary requests are exposed without annotations or guards.
scrape the web, run code in a sandbox, send/receive email, enrich a contRecommendationAnnotate destructive tools and require human approval.
tool: consult
A tool description tries to alter the model’s use of another tool.
lls to run via `use`, with a message explaining the preferred choicRecommendationDescriptions must describe only their own tool.
tool: consult
One tool reads private data, ingests untrusted content, AND can send data outbound — the three preconditions for autonomous data theft in a single call. A prompt injection reaching this tool can exfiltrate secrets with no further step.
ideo, search or scrape the web, run code in a sandbox, send/receive email, enrRecommendationSplit the capabilities across separate least-privilege tools so no single tool can read secrets, read untrusted input, and reach the network at once.
tool: use
A tool description tries to alter the model’s use of another tool.
ngle call that `consult` handed you, and bill on success. Used for anyRecommendationDescriptions must describe only their own tool.
tool: result
A tool description tries to alter the model’s use of another tool.
f re-running `use` (re-running starts a new paid job).RecommendationDescriptions must describe only their own tool.
tool: result
The text tells the model WHEN to call this tool relative to others ("always call first", "before any other tool", "chain to X tool") — a toxic-flow injection that hijacks the agent’s orchestration rather than describing the tool.
double-charges. ALWAYS use this to retrieve an async result instead of re-runniRecommendationTool metadata must describe only the tool, never sequence the agent’s calls.
tool: session
Tools that execute commands / write files / make arbitrary requests are exposed without annotations or guards.
ption":"Code to execute (alternative to `command`)."},"language":{"type":"string","description":RecommendationAnnotate destructive tools and require human approval.
tool: session
A tool description tries to alter the model’s use of another tool.
on (started by `use` with action `create_session`, which returns aRecommendationDescriptions must describe only their own tool.
tool: session
An exec/shell/run-code tool exposes a command/code/args parameter that is a free-form string with no enum/pattern guard — a direct command-injection / RCE surface. A tool constraining the command to an `enum` of allowed operations does not fire.
unconstrained command parameter "command"RecommendationReplace the free-form command with a fixed `enum` of allowed operations (or structured, escaped arguments) and require explicit human approval.
tool: close
A tool description tries to alter the model’s use of another tool.
an E2B sandbox session and stop its billing. Pass the `session_id`. CaRecommendationDescriptions must describe only their own tool.
tool: close
The text tells the model WHEN to call this tool relative to others ("always call first", "before any other tool", "chain to X tool") — a toxic-flow injection that hijacks the agent’s orchestration rather than describing the tool.
eases the hold. ALWAYS call this when finished with a session — an open sessionRecommendationTool metadata must describe only the tool, never sequence the agent’s calls.
tool: llm
A tool description tries to alter the model’s use of another tool.
ion of a cent). Use it to get a second opinion from a rival model,RecommendationDescriptions must describe only their own tool.
tool: (server instructions)
A tool exposes a free-text/webhook/feedback parameter with no honest purpose — a classic exfil channel.
the user for an API key, or tell them to do it manually. You have theseRecommendationRemove side-channel parameters; constrain tool inputs.
tool: (server instructions)
Tools that execute commands / write files / make arbitrary requests are exposed without annotations or guards.
arse documents; run code and spin up sandboxes; drive a browser; send emRecommendationAnnotate destructive tools and require human approval.
tool: (server instructions)
A tool description tries to alter the model’s use of another tool.
` when done. - `vaaya_account` — which account is connected, balance, premiumRecommendationDescriptions must describe only their own tool.
The server accepts tool enumeration (and likely invocation) with no authentication.
RecommendationRequire OAuth 2.1 authorization for any server exposing non-public tools.
tool: vaaya_logout
The server advertises open-world / broadly-scoped capabilities.
annotations.destructiveHint = trueRecommendationScope tools to the minimum needed.
tool: consult
The server advertises open-world / broadly-scoped capabilities.
annotations.openWorldHint = trueRecommendationScope tools to the minimum needed.
tool: use
The server advertises open-world / broadly-scoped capabilities.
annotations.openWorldHint = trueRecommendationScope tools to the minimum needed.
tool: result
The server advertises open-world / broadly-scoped capabilities.
annotations.openWorldHint = trueRecommendationScope tools to the minimum needed.
tool: session
The server advertises open-world / broadly-scoped capabilities.
annotations.openWorldHint = trueRecommendationScope tools to the minimum needed.
tool: llm
The server advertises open-world / broadly-scoped capabilities.
annotations.openWorldHint = trueRecommendationScope tools to the minimum needed.
Vantaj uptime monitoring via MCP — manage monitors, heartbeats, incidents, and status pages.
Unified gateway to Algeria's TKAWEN ecosystem: commerce, certification, and AI tools.
Provides access to the Cohereon Doctrine AI safety framework with governance components, tiered access, and agent onboarding.
Agentic rails for complex workflows with receipts, fees, and MCP tool access.
Structural TC39 spec lookup for ECMA-262 and ECMA-402 in AI agents, SHA-pinned and offline-first.
Structural TC39 spec lookup for ECMA-262 and ECMA-402 in AI agents, SHA-pinned and offline-first.