io.github.Areso/safe-ssh-mcp

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

  pass25% of grade

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

  Scanned 224 text field(s); no instruction-injection patterns found. PASS = no static red flag. A server can serve different tool descriptions at runtime (the rug-pull class), which static scanning cannot see.

• 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

  fail15% of grade

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

  tool "get_disk_free" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_inode_usage" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_disk_usage" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_list_of_files" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_list_of_files_with_filter" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_is_installed" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_dmesg" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_uptime" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_current_datetime" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_distroname_version" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_systemd_list_all" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_systemd_list_failed" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_systemd_list_timers" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_systemd_status" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_ps_aux_top_cpu_consumers" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_ps_aux_top_mem_consumers" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_free_memory" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_memory_pressure" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_top" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_lsblk" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_docker_ps_all" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_crontab_tasks" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_listening_sockets" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "get_service_logs_from_journalctl" has a secret-shaped input parameter "password" — 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

  fail10% of grade

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

  vs prior 0.1.14 (patch bump): new tool "get_disk_free" appeared; new tool "get_inode_usage" appeared; new tool "get_disk_usage" appeared; new tool "get_list_of_files" appeared; new tool "get_list_of_files_with_filter" appeared; new tool "get_is_installed" touches a sensitive surface (email/exec/net/fs); new tool "get_dmesg" appeared; new tool "get_uptime" appeared; new tool "get_current_datetime" appeared; new tool "get_distroname_version" appeared; new tool "get_systemd_list_all" appeared; new tool "get_systemd_list_failed" appeared; new tool "get_systemd_list_timers" appeared; new tool "get_systemd_status" appeared; new tool "get_ps_aux_top_cpu_consumers" appeared; new tool "get_ps_aux_top_mem_consumers" appeared; new tool "get_free_memory" appeared; new tool "get_memory_pressure" appeared; new tool "get_top" appeared; new tool "get_lsblk" appeared; new tool "get_docker_ps_all" appeared; new tool "get_crontab_tasks" appeared; new tool "get_listening_sockets" appeared; new tool "get_service_logs_from_journalctl" appeared.

• Transport config

  unscannable10% of grade

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

  local stdio server: no remote transport or auth surface, and no phone-home/telemetry signal found in source. Absence of transport findings is NOT a safety guarantee — phone-home hosts can be constructed at runtime and defeat static analysis.