Security: harden the exec/FS trust boundary in server_actions

[hyperpolymath]

Summary

Closes three injection classes reachable from GUI-supplied values through src/interface/ffi/src/server_actions.zig. Part of the v1.0 hardening roadmap (follows #57/#58).

Changes

• Remote command injection (primary). The SSH branches in executeAction/streamLogs/getServerStatus built ssh <host> podman … <container_name> with no protection. SSH concatenates its post-target args into one string that the remote login shell re-parses, so a container_name like foo; rm -rf / executed the rm. The pre-existing executeSSH helper was unused and its -- does not stop remote-shell reparsing (its doc comment claimed otherwise — corrected).
• SSH option injection. An unvalidated host such as -oProxyCommand=evil is read by ssh as an option.
• Path traversal / XML injection. run_script accepted scripts/../../etc/x (prefix check only); write_server_config interpolated a raw profile_id into a filesystem path and wrote Settings.xml element text with no escaping (operator attributes escaped only ").

Fixes:

• Allowlist validators isSafeContainerName / isSafeHost / isSafeProfileId reject every shell metacharacter and path-traversal character at the boundary.
• Single SSH path: callers build a bare command and route through dispatch() → executeSSH, which now POSIX-single-quotes every remote arg (shellQuoteInto) as defence-in-depth.
• run_script also rejects .. components and absolute paths; write_server_config validates profile_id and XML-escapes all five predefined entities in every interpolated value.
• parseAndDispatch converts validation errors into a structured failure result instead of propagating.
• Also adds timeout-minutes to push-email-notify.yml (clears the Hypatia workflow_audit finding).

RSR Quality Checklist

Required

• Tests pass (zig build test{,-integration,-smoke,-property} — 150 green: 92 unit / 39 integration / 5 smoke / 14 property)
• No banned language patterns (Zig only)
• No banned functions introduced
• SPDX license headers present on modified source files
• No secrets, credentials, or .env files included

As Applicable

• ABI/FFI changes validated (no ABI surface change; abi-contract.yml gates unaffected)
• Documentation updated for user-facing changes (n/a — internal hardening)

Testing

10 new adversarial unit tests in server_actions.zig: injection payloads (;, backticks, $(…), pipes) rejected by the validators; -oProxyCommand host rejected; scripts/../../etc/passwd traversal rejected; shellQuoteInto/xmlEscapeInto output asserted; executeAction/run_script/write_server_config reject malicious inputs at the boundary. Grep-gate: no inline ssh arg-building remains.

🤖 Generated with Claude Code

https://claude.ai/code/session_01TaWWedv6VQqeZaPvc94keN

---

Generated by Claude Code

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[github-actions]

🔍 Hypatia Security Scan

Findings: 2 issues detected

Severity	Count
🔴 Critical	0
🟠 High	0
🟡 Medium	2

View findings

[
  {
    "reason": "Repository has 21 non-main remote branch(es). Policy: single main branch only.",
    "type": "GS007",
    "file": ".",
    "action": "delete_remote_branches",
    "rule_module": "git_state",
    "severity": "medium"
  },
  {
    "reason": "Code scanning (Hypatia): hypatia/workflow_audit/missing_timeout_minutes -- Hypatia workflow_audit: missing_timeout_minutes -- 7 day(s) old",
    "type": "CSA001",
    "file": "push-email-notify.yml",
    "action": "review",
    "rule_module": "code_scanning_alerts",
    "severity": "medium"
  }
]

Powered by Hypatia Neurosymbolic CI/CD Intelligence