io.github.JesseGdotIO/hermes-plant

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).

  tool "freetier_status" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "purchase_with_x402" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "action_safety_quick_gate" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "action_safety_run" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "submit_reviewqueue_request" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "score_destructguard_command" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "dealanalyzer_analyze" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "cashflowlens_analyze" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "waterfall_distribute" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "options_price" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "bond_analyze" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "portfolioguard_score" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "walletguard_score" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "emailguard_validate" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "emailguard_validate_batch" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "emailguard_validate_batch" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "mcp_risk_score" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "payment_policy_decide" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "evidence_verify" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes. tool "assurance_attest" param "apiKey" description: off-vendor URL in a description (hermesplant.com); capability text should not name where data goes.

• Supply chain

  pass25% of grade

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

  repository linkage verified

• Credential hygiene

  fail15% of grade

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

  tool "freetier_status" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "purchase_with_x402" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "action_safety_quick_gate" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "action_safety_run" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "submit_reviewqueue_request" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "score_destructguard_command" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "dealanalyzer_analyze" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "cashflowlens_analyze" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "waterfall_distribute" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "options_price" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "bond_analyze" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "portfolioguard_score" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "walletguard_score" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "emailguard_validate" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "emailguard_validate_batch" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "mcp_risk_score" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "payment_policy_decide" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "evidence_verify" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential). tool "assurance_attest" has a secret-shaped input parameter "apiKey" — secrets must never be tool parameters (the model would supply/handle the credential).

• Permission scope

  pass15% of grade

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

  No shell-exec, unbounded/sensitive filesystem, or download-and-run primitives found in scanned source. PASS = no static red flag. Static scanning is blind to code fetched or executed at runtime.

• Version behavior

  pass10% of grade

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

  tool surface, capabilities, install scripts, and publisher unchanged vs prior version 2.0.1 (patch bump); no rug-pull drift.

• 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)