io.github.adewale/agentic-mermaid

What we checked

Six static checks, weighted by risk. Every result reflects only what could be observed in the published package and repository — never intent.

• Injection surface

  warn25% of grade

  Tool descriptions/manifest scanned for instruction-injection patterns (imperatives at the model, hidden text, 'ignore previous', data-exfil URLs).

  server instructions: off-vendor URL in a description (agentic-mermaid.dev); capability text should not name where data goes.

• Supply chain

  warn25% of grade

  Package provenance: namespace verification, repo linkage, maintainer count, account age, postinstall scripts, typosquat distance.

  dependency "yaml" is edit-distance 2 from popular package "yarn" (typosquat candidate); dependency "elkjs" is edit-distance 2 from popular package "ejs" (typosquat candidate)

• Credential hygiene

  pass15% of grade

  How the server takes secrets (env vs plaintext config vs hardcoded); secrets appearing in tool schemas.

  Reads secrets from the environment (e.g. src/cli/index.ts) — the recommended intake shape. PASS = no static red flag; static analysis cannot prove the code honors env-based secret handling at runtime.

• Permission scope

  warn15% of grade

  Declared tools vs. breadth (filesystem, network, exec); flags shell-exec and unbounded filesystem access.

  has both shell/exec and network egress — capable of download-and-run, though no such path was observed (examples/agent-improve-auth-flow.ts: imports child_process and calls exec/spawn; src/mcp/dynamic-harness.ts: outbound HTTP/socket call). discloses shell/exec capability: examples/agent-improve-auth-flow.ts: imports child_process and calls exec/spawn discloses shell/exec capability: src/cli/index.ts: imports child_process and calls exec/spawn discloses shell/exec capability: examples/mcp-vs-cli-complex-diagrams.ts: imports child_process and calls exec/spawn discloses dynamic code execution: src/mcp/facade.ts: uses new Function(…) Capability DISCLOSURE, not a verdict: static analysis sees the primitive is present and reachable, not whether its use is attacker-controlled.

• Version behavior

  fail10% of grade

  Diff of tool definitions between versions; new permissions or changed descriptions in a patch release (the postmark-mcp class).

  vs prior 0.4.0 (patch bump): new capability primitive(s) in source: child_process, dynamic-eval, http-client; new hardcoded outbound host/BCC destination(s): agentic-mermaid.dev, fonts.googleapis.com, github.com, llmstxt.org, www.w3.org.

• Transport config

  fail10% of grade

  Remote servers: TLS and auth mode (none/token/OAuth). Local servers: whether the manifest indicates it phones home.

  remote MCP endpoint declares NO authentication (unauthenticated remote — nginx-ui/CVE-2026-33032 class); manifest/source references tracking behavior — disclosure