Strix (STRIX) — interface

docs / modules / interface

← all docs

Strix (STRIX) — interface

flowchart LR IN["IN: strix --target ...\n(argv)"] --> ACT["ACT: parse args, prep run,\ndispatch to headless CLI\nor interactive TUI"] ACT --> VAL["VAL: Docker preflight,\nLLM connectivity warm-up"] VAL --> OUT["OUT: run_strix_scan() invoked;\nlive display + final artifacts"]

What this is

strix/interface is everything a human (or a CI job) actually touches: the command line, the live interactive terminal view, and the local web dashboard for browsing finished runs. It doesn't do any testing itself — its whole job is to collect arguments, hand off to strix/core, and render what comes back.

What flows through it

flowchart TD Main["main.py\nmain()"] --> Args["cli_args.py\nparse_arguments()"] Main --> Setup["scan_setup.py\nprepare_run()"] Main -- "-n / --non-interactive" --> CLI["cli.py\nrun_cli() -> run_strix_scan()"] Main -- "default" --> Interactive["interactive.py\nrun_tui() -> run_strix_scan()"] Interactive --> TUI["tui/\nGo Bubble Tea frontend\n<-> Python backend server"] CLI -.-> Viewer["viewer/\nstrix view — local\nread-only web dashboard"]

Technical details

flowchart LR IN["IN: strix --target ...\n(argv)"] --> ACT["ACT: parse args, prep run,\ndispatch to headless CLI\nor interactive TUI"] ACT --> VAL["VAL: Docker preflight,\nLLM connectivity warm-up"] VAL --> OUT["OUT: run_strix_scan() invoked;\nlive display + final artifacts"]

Package: strix/interface/ Depends on: strix.core.runner (run_strix_scan, the sole scan entry point), strix.config, strix.report.state. Depended on by: nothing inside strix/ — this is the outermost layer, the process entry point (strix console script).

Files at a glance

File Role
main.py main() — top-level dispatch: strix view, strix auth, or a scan.
cli_args.py parse_arguments() — the full argparse surface (--target, --instruction, --scan-mode, --max-budget-usd, --max-turns, --resume, -n, etc.).
scan_setup.py prepare_run() — resolves run_name, clones repo targets, computes local_sources/API-spec staging/diff-scope, persists the initial run record.
cli.py run_cli() — headless mode: builds scan_config, sets up ReportState, calls run_strix_scan() under a Rich Live status panel, prints the final report, exits 2 if findings exist.
interactive.py run_tui() — async: launches and supervises the Go TUI, itself calling run_strix_scan().
auth_cli.py strix auth login/status/logout — ChatGPT-subscription auth flow (_CodexResponsesModel in config/models.py).
environment.py validate_environment(), warm_up_llm() — preflight checks (Docker installed, model connectivity) run before a scan starts.
update_check.py Background version-check + update prompt.
tui/ Go Bubble Tea frontend (cmd/strix-tui) + a Python IPC backend (backend/server.py, controller.py, protocol.py) streaming live agent-tree/tool-call/vulnerability events. (Go internals not read line-by-line — role inferred from file/directory structure.)
viewer/ strix view — a stdlib ThreadingHTTPServer (server.py) serving a prebuilt React/Vite SPA and JSON endpoints that read strix_runs/<run>/ straight off disk (transcript.py), plus auth.py (local token gating) and report_pdf.py (PDF export).

Files & functions, explained

Flow from strix --target X to a running scan (see ../design/data-flow.md for the full numbered sequence): main() → parse_arguments() → Docker preflight/image pull → _bootstrap_scan(args) (validate_environment(), warm_up_llm() for non-interactive runs, persist_current() writes resolved config back to ~/.strix/cli-config.json, scan_setup.prepare_run(args)) → dispatch to cli.run_cli(args) or interactive.run_tui(args) → both ultimately call core.runner.run_strix_scan(...).

Related: STRIX — Architecture (HLD), core.md, ../design/data-flow.md, ../process/runbooks/run-a-detection-scan.md.