io.github.mlawsonking/agent-guards

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

  unscannable25% of grade

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

  No tool descriptions, server instructions, prompts, or resources were fetched; nothing to scan for injection.

• Supply chain

  pass25% of grade

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

  repository linkage verified

• 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. cli/exec.js) — 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 (cli/exec.js: imports child_process and calls exec/spawn; engines/email.js: outbound HTTP/socket call). has both shell/exec and destructive filesystem writes (cli/exec.js: imports child_process and calls exec/spawn; lib/cache.js: destructive/writing fs.* call (write/unlink/rm/chmod/rename)). discloses shell/exec capability: cli/exec.js: imports child_process and calls exec/spawn discloses dynamic code execution: engines/code.js: uses new Function(…) discloses dynamic code execution: engines/code.js: uses eval(…) discloses broad/destructive filesystem access: lib/cache.js: destructive/writing fs.* call (write/unlink/rm/chmod/rename) discloses broad/destructive filesystem access: lib/feed.js: destructive/writing fs.* call (write/unlink/rm/chmod/rename) 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 0.2.1 (minor bump): new capability primitive(s) in source: child_process, dynamic-eval, fs-write, http-client, subprocess; new hardcoded outbound host/BCC destination(s): 1rpc.io, agent-firewall-seven.vercel.app, api.honeypot.is, api.npmjs.org, api.osv.dev, api.pwnedpasswords.com, arb1.arbitrum.io, arbitrum-one-rpc.publicnode.com, base-rpc.publicnode.com, check.torproject.org, eth.llamarpc.com, ethereum-rpc.publicnode.com, ftp, github.com, mainnet.base.org, mainnet.optimism.io, optimism-rpc.publicnode.com, polygon-bor-rpc.publicnode.com, polygon-rpc.com, pypi.org, pypistats.org, raw.githubusercontent.com, rdap.org, registry.npmjs.org.

• Transport config

  warn10% of grade

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

  manifest/source references tracking behavior — disclosure