io.github.abe238/hotin

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.

  Python build script has no build-time network/exec; 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. src/hotin/cache.py) — 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 (src/hotin/mcp.py: subprocess.* call; src/hotin/sources/_hf.py: outbound HTTP/socket call). has both shell/exec and destructive filesystem writes (src/hotin/mcp.py: subprocess.* call; src/hotin/config.py: destructive filesystem call (shutil.rmtree/os.remove/etc.)). discloses shell/exec capability: src/hotin/mcp.py: subprocess.* call discloses shell/exec capability: src/hotin/schedule.py: subprocess.* call discloses broad/destructive filesystem access: src/hotin/config.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) 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 0.9.3 (patch bump): new capability primitive(s) in source: http-client, subprocess; new hardcoded outbound host/BCC destination(s): a.test, acme.example, ainativefoundation.org, api.github.com, api.npmjs.org, api.ossinsight.io, api.scrapecreators.com, arxiv.org, b.test, bad.example, blog.google, deepmind.google, evil.example, example.com, github.com, hn.algolia.com, hotin.ai, huggingface.co, img.shields.io, importai.substack.com, mistral.ai, news.smol.ai, openai.com, qwenlm.github.io, registry.npmjs.org, semianalysis.com, simonwillison.net, thinkingmachines.ai, www.example.com, www.googleapis.com, www.interconnects.ai, www.latent.space, www.microsoft.com, www.reddit.com, www.rohan-paul.com, x, x.test.

• 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