[SizzleUnrlsd]

Summary

Keeps the rule id command-injection for every case, so existing suppressions and baselines keep matching (#220). When the rule reads the command element by element and establishes what the attacker's input is, the finding carries an injection metadata field:

• command: the input runs or chooses what runs (shell string, program, operand run as a command, value of an option that runs it); title Command injection;
• option: the input can only reach the program as an option or as the value of an option not established to run it (cases 4, 5b, 5c, 7 of command-injection: severity of option injection in argument lists run without a shell #208); title Option injection.

When the kind cannot be established (an option whose effect is to review, an unknown program or element before the input, a command that is not a list or tuple display, a sink such as os.system not read element by element), the field is absent and the title stays Command injection.

Assessment gains optional title and metadata fields (plugin API, backward compatible). Documented in docs/usage.md and docs/plugins.md. The SARIF report does not carry finding metadata today, so the field appears in the JSON report.

Verification

New tests in tests/test_command_injection_nature.py (red before the change), including an inline ignore[command-injection] suppressing an option injection. No existing test changed. Full suite green apart from the local test_packaging environment issue; ruff, mypy clean; regression snapshots unchanged; mitmproxy scan: same 75 findings (rule, severity, confidence, location), titles of three option cases changed.

Closes #220