app.agentbit/mcp v1.3.0
Host app.agentbit/mcp in a governed environment — SSO in front, every tool call attributed and audited.
Security controls
Deterministic, evidence-backed. Score earned from passing controls; a failed guard caps it.
Security controls
Deterministic, evidence-backed. Score earned from passing controls; a failed guard caps it.
Model–MCP Runtime Guardrails
- WarnStrict JSON Schema Enforcementschemas not strict
Tool inputs are constrained (additionalProperties:false), so unexpected arguments can't be smuggled in.
Fix: Set additionalProperties:false and require explicit parameters on every tool.
- PassIndirect Prompt Injection (IPI) Defensesguardno injection markers
Tool/prompt/resource text is free of hidden instructions that could hijack the agent.
- PassTool Definition Integrityguardtool definitions stable; no risky changes in version history
Every consecutive version pair is diffed for new injection markers or destructive scope — a risky diff anywhere in history is a rug-pull (fail, durable); benign description drift warns.
- PassIn-Session Definition Integrityguardno call-gated or call-time tool definition changes
Tool and prompt definitions stay fixed during a session — they don't change after a few calls, which is how runtime-gated rug-pulls slip past one-time review.
- PassUser-in-the-Loop / Approval Scopeguardno destructive tool scope
No over-broad or destructive tools (arbitrary shell, bulk-delete) that warrant human approval.
Application Security Checks
- WarnCredentials sourced from environmentno env-based config detected
Reading secrets from env/secret stores avoids hardcoding them.
Fix: Read credentials from environment variables or a secret manager.
- WarnDependencies pinned (lockfile)no lockfile found
A lockfile makes installs reproducible and resistant to silent dependency swaps.
Fix: Commit a lockfile (package-lock.json / pnpm-lock.yaml / poetry.lock).
- WarnHas a security policyno security policy
A SECURITY.md gives a private path to report vulnerabilities.
Fix: Add SECURITY.md with a disclosure contact and process.
- WarnAdoption & popularitylimited adoption
A small, capped nudge from stars/downloads — widely-used servers get more eyes on bugs. It can never offset a real security failure.
Fix: Adoption grows naturally with real-world use; this never gates the score.
- PassNo command-injection sinksguardno sinks found
Untrusted tool input reaching a shell yields remote code execution.
- PassNo dynamic code executionguardno sinks found
eval()/exec()/Function() on tool-derived strings allows arbitrary code execution.
- PassNo path traversalguardno sinks found
Naive path checks let tools read/write outside intended directories (EscapeRoute-class).
- PassNo SSRF sinksguardno sinks found
Fetching tool-supplied URLs can pivot into internal networks and metadata services.
- PassNo unsafe deserializationguardno sinks found
pickle/yaml.load/etc. on untrusted data can execute code.
- PassNo committed secretsguardno secrets found
Hardcoded keys/tokens in published source are live credentials an attacker can use.
- PassActively maintainedrecent commits
Unmaintained servers don't receive security fixes.
- PassRepository not archivedguardactive
Archived repositories will never be patched.
- PassDeclares a licenseMIT
A clear license is required for legal enterprise use.
- Not checkedNo known-vulnerable dependenciesno lockfile found
Runtime dependencies (parsed from the lockfile) are scanned against OSV.dev for published CVEs. Advisory: flagged dependencies lower the score but don't hard-cap it, since transitive reachability is unproven.
- Not checkedSigned releasesnot evaluated
Signed releases let consumers verify artifacts weren't tampered with.
- N/ANo install/post-install scriptsguardno published package
install hooks run arbitrary code on every consumer at install time.
- N/APackage name not typosquattingguardno published package
Names mimicking popular packages are a common malware delivery vector.
- N/APublished with provenanceno published package
Build provenance attests the artifact was built from the claimed source by CI.
- N/AEstablished maintainerno published package
Brand-new / single anonymous maintainers raise takeover and malware risk.
Transport & Trust Model
- WarnIAM / Authentication Scopingno authentication handling detected
OAuth 2.1 / Protected Resource Metadata gates who can invoke tools.
Fix: Implement OAuth 2.1 with a .well-known/oauth-protected-resource document.
- WarnExecution Sandboxingruns natively (no container image)
A container/sandbox image limits blast radius; a server that runs natively has full host access.
Fix: Ship a Dockerfile/Containerfile (or document a sandboxed run) so the server runs isolated.
- PassTransport Encryption (TLS)guardall remotes use HTTPS
Plaintext HTTP exposes traffic and bearer tokens to interception.
- PassNo Token Passthroughguardno inbound token forwarded downstream
The client's token is never forwarded to downstream APIs — the server obtains its own, correctly-scoped credential instead.
- PassNetwork Exposureno bind-all detected
Binding 0.0.0.0 or exposing debug inspectors widens the attack surface.
- Not checkedDNS-Rebinding ProtectionHTTP transport not found in source
HTTP endpoints reject requests carrying an unlisted Origin, so a malicious web page can't drive the server through the user's browser.
- N/ALive Endpoint Reachablenot dynamically scanned
A dynamic scan connected to the declared remote endpoint and it responded — verified live, not a dead URL.
- N/AAuthentication Enforced (live)not dynamically scanned
If the server declares auth is required, it must actually reject anonymous clients. Serving tools to unauthenticated callers is a real exposure.