Skip to main content

hematite/agent/
routing.rs

1use super::conversation::WorkflowMode;
2
3#[derive(Clone, Copy, Debug, Eq, PartialEq)]
4pub(crate) enum QueryIntentClass {
5    ProductTruth,
6    RuntimeDiagnosis,
7    RepoArchitecture,
8    Toolchain,
9    Capability,
10    Implementation,
11    Unknown,
12}
13
14#[derive(Clone, Copy, Debug, Eq, PartialEq)]
15pub(crate) enum DirectAnswerKind {
16    About,
17    LanguageCapability,
18    UnsafeWorkflowPressure,
19    SessionMemory,
20    RecoveryRecipes,
21    McpLifecycle,
22    AuthorizationPolicy,
23    ToolClasses,
24    ToolRegistryOwnership,
25    SessionResetSemantics,
26    ProductSurface,
27    ReasoningSplit,
28    Identity,
29    WorkflowModes,
30    GemmaNative,
31    GemmaNativeSettings,
32    VerifyProfiles,
33    Toolchain,
34    ArchitectSessionResetPlan,
35}
36
37#[derive(Clone, Copy, Debug)]
38pub(crate) struct QueryIntent {
39    pub(crate) primary_class: QueryIntentClass,
40    pub(crate) direct_answer: Option<DirectAnswerKind>,
41    pub(crate) grounded_trace_mode: bool,
42    pub(crate) capability_mode: bool,
43    pub(crate) capability_needs_repo: bool,
44    pub(crate) toolchain_mode: bool,
45    pub(crate) host_inspection_mode: bool,
46    pub(crate) maintainer_workflow_mode: bool,
47    pub(crate) workspace_workflow_mode: bool,
48    pub(crate) architecture_overview_mode: bool,
49}
50
51fn contains_any(haystack: &str, needles: &[&str]) -> bool {
52    needles.iter().any(|needle| haystack.contains(needle))
53}
54
55fn contains_all(haystack: &str, needles: &[&str]) -> bool {
56    needles.iter().all(|needle| haystack.contains(needle))
57}
58
59fn mentions_reset_commands(lower: &str) -> bool {
60    contains_all(lower, &["/clear", "/new", "/forget"])
61}
62
63fn mentions_stable_product_surface(lower: &str) -> bool {
64    contains_any(
65        lower,
66        &[
67            "stable product-surface question",
68            "stable product surface question",
69            "stable product-surface questions",
70            "stable product surface questions",
71        ],
72    )
73}
74
75fn mentions_product_truth_routing(lower: &str) -> bool {
76    let asks_decision_policy = contains_any(
77        lower,
78        &[
79            "how hematite decides",
80            "how does hematite decide",
81            "decides whether",
82            "decide whether",
83        ],
84    );
85    let asks_direct_vs_inspect_split = contains_any(
86        lower,
87        &[
88            "answered as stable product truth",
89            "stable product truth",
90            "stable product behavior",
91            "answer directly",
92            "direct answer",
93            "inspect the repository",
94            "inspect repository",
95            "repository implementation",
96            "repo implementation",
97        ],
98    );
99    asks_decision_policy && asks_direct_vs_inspect_split
100}
101
102fn mentions_broad_system_walkthrough(lower: &str) -> bool {
103    let asks_walkthrough = contains_any(
104        lower,
105        &[
106            "walk me through",
107            "walk through",
108            "how hematite is wired",
109            "understand how hematite is wired",
110            "major runtime pieces",
111            "normal message moves",
112            "moves from the tui to the model and back",
113        ],
114    );
115    let asks_multiple_runtime_areas = contains_any(
116        lower,
117        &[
118            "session recovery",
119            "tool policy",
120            "mcp state",
121            "mcp policy",
122            "files own the major runtime pieces",
123            "which files own",
124            "where session recovery",
125            "where tool policy",
126            "where mcp state",
127        ],
128    );
129    asks_walkthrough && asks_multiple_runtime_areas
130}
131
132fn mentions_capability_question(lower: &str) -> bool {
133    contains_any(
134        lower,
135        &[
136            "what can you do",
137            "what are you capable",
138            "can you make projects",
139            "can you build projects",
140            "do you know other coding languages",
141            "other coding languages",
142            "what languages",
143            "can you use the internet",
144            "internet research capabilities",
145            "what tools do you have",
146        ],
147    )
148}
149
150fn mentions_creator_question(lower: &str) -> bool {
151    contains_any(
152        lower,
153        &[
154            "who created you",
155            "who built you",
156            "who made you",
157            "who developed you",
158            "who engineered you",
159            "who engineered your architecture",
160            "who created hematite",
161            "who built hematite",
162            "who developed hematite",
163            "who engineered hematite",
164            "who maintains hematite",
165            "who authored hematite",
166            "who is the author",
167            "who wrote this",
168            "who made this app",
169        ],
170    )
171}
172
173fn capability_question_requires_repo_inspection(lower: &str) -> bool {
174    contains_any(
175        lower,
176        &[
177            "this repo",
178            "this repository",
179            "codebase",
180            "which files",
181            "implementation",
182            "in this project",
183        ],
184    )
185}
186
187fn mentions_host_inspection_question(lower: &str) -> bool {
188    let host_scope = contains_any(
189        lower,
190        &[
191            "path",
192            "package manager",
193            "package managers",
194            "env doctor",
195            "environment doctor",
196            "pip",
197            "winget",
198            "choco",
199            "scoop",
200            "network",
201            "adapter",
202            "dns",
203            "gateway",
204            "ip address",
205            "ipconfig",
206            "wifi",
207            "ethernet",
208            "service",
209            "services",
210            "daemon",
211            "startup type",
212            "process",
213            "processes",
214            "task manager",
215            "ram",
216            "cpu",
217            "memory",
218            "developer tools",
219            "toolchains",
220            "installed",
221            "desktop",
222            "downloads",
223            "folder",
224            "directory",
225            "local development",
226            "machine",
227            "computer",
228            "firewall",
229            "vpn",
230            "proxy",
231            "internet",
232            "online",
233            "connectivity",
234            "ssid",
235            "wireless",
236            "tcp connection",
237            "active connection",
238            "traceroute",
239            "tracert",
240            "dns cache",
241            "arp table",
242            "arp cache",
243            "route table",
244            "routing table",
245            "default gateway",
246            "next hop",
247            "power plan",
248            "power settings",
249            "uptime",
250            "reboot",
251            "health",
252            "report",
253            "bitlocker",
254            "rdp",
255            "remote desktop",
256            "vss",
257            "shadow copy",
258            "shadow copies",
259            "pagefile",
260            "virtual memory",
261            "swap",
262            "windows feature",
263            "optional feature",
264            "printer",
265            "print queue",
266            "winrm",
267            "psremoting",
268            "network stats",
269            "adapter stats",
270            "udp listening",
271            "udp port",
272            "session",
273            "logon",
274            "login",
275            "virtualization",
276            "hypervisor",
277            "vt-x",
278            "slat",
279        ],
280    );
281    let host_action = contains_any(
282        lower,
283        &[
284            "inspect",
285            "count",
286            "tell me",
287            "summarize",
288            "how big",
289            "biggest",
290            "versions",
291            "duplicate",
292            "missing",
293            "ready",
294            "fix",
295            "repair",
296            "resolve",
297            "troubleshoot",
298        ],
299    );
300
301    host_scope && host_action
302}
303
304pub fn preferred_host_inspection_topic(user_input: &str) -> Option<&'static str> {
305    let lower = user_input.to_lowercase();
306    let asks_fix_plan = (lower.contains("fix")
307        || lower.contains("repair")
308        || lower.contains("resolve")
309        || lower.contains("troubleshoot"))
310        && (lower.contains("cargo")
311            || lower.contains("path")
312            || lower.contains("package manager")
313            || lower.contains("toolchain")
314            || lower.contains("port ")
315            || lower.contains("already in use")
316            || lower.contains("lm studio")
317            || lower.contains("localhost:1234")
318            || lower.contains("embedding model")
319            || lower.contains("no coding model loaded"));
320    let asks_path = lower.contains("path entries")
321        || lower.contains("raw path")
322        || (lower.contains("path") && (lower.contains("show") || lower.contains("what is")));
323    let asks_gpo = lower.contains("gpo")
324        || lower.contains("group policy")
325        || lower.contains("gpresult")
326        || lower.contains("applied policy");
327    let asks_certificates = lower.contains("cert")
328        || lower.contains("ssl")
329        || lower.contains("client cert")
330        || lower.contains("expiring cert");
331    let asks_integrity = lower.contains("integrity")
332        || lower.contains("sfc")
333        || lower.contains("dism")
334        || lower.contains("corruption")
335        || lower.contains("os health");
336    let asks_domain = lower.contains("domain")
337        || lower.contains("active directory")
338        || lower.contains("ad join")
339        || lower.contains("workgroup")
340        || lower.contains("netbios");
341    let asks_device_health = lower.contains("device health")
342        || lower.contains("hardware error")
343        || lower.contains("malfunctioning")
344        || lower.contains("yellow bang")
345        || lower.contains("hardware failing");
346    let asks_drivers =
347        lower.contains("driver") || lower.contains("kmod") || lower.contains("kernel module");
348    let asks_peripherals = lower.contains("peripheral")
349        || lower.contains("usb")
350        || lower.contains("keyboard")
351        || lower.contains("mouse")
352        || lower.contains("pointer")
353        || lower.contains("monitor")
354        || lower.contains("input device")
355        || lower.contains("connected hardware");
356    let asks_sessions = lower.contains("session")
357        || lower.contains("login")
358        || lower.contains("who is on")
359        || lower.contains("active user");
360    let asks_virtualization = lower.contains("virtualization")
361        || lower.contains("hypervisor")
362        || lower.contains("vt-x")
363        || lower.contains("slat")
364        || lower.contains("v-p")
365        || lower.contains("nested virt")
366        || lower.contains("cpu model")
367        || lower.contains("ram size")
368        || lower.contains("hardware spec")
369        || lower.contains("hardware dna")
370        || lower.contains("hardware info")
371        || lower.contains("bios version")
372        || lower.contains("motherboard")
373        || lower.contains("gpu name")
374        || lower.contains("gpu driver")
375        || lower.contains("how much ram")
376        || lower.contains("how much vram")
377        || lower.contains("what processor")
378        || lower.contains("what cpu")
379        || lower.contains("what gpu")
380        || (lower.contains("what hardware") && lower.contains("have"))
381        || (lower.contains("hardware") && lower.contains("inventory"));
382    let asks_startup = lower.contains("startup")
383        || lower.contains("boot program")
384        || lower.contains("autorun")
385        || lower.contains("run at boot");
386    let asks_env_doctor = lower.contains("env doctor")
387        || lower.contains("environment doctor")
388        || lower.contains("package manager")
389        || lower.contains("package managers")
390        || lower.contains("shims")
391        || lower.contains("path drift")
392        || (lower.contains("dev machine") && lower.contains("off"))
393        || (lower.contains("environment") && lower.contains("sane"));
394    let asks_network = lower.contains("network")
395        || lower.contains("adapter")
396        || lower.contains("dns")
397        || lower.contains("gateway")
398        || lower.contains("ip address")
399        || lower.contains("ipconfig")
400        || lower.contains("wifi")
401        || lower.contains("ethernet")
402        || lower.contains("subnet");
403    let asks_services = lower.contains("service")
404        || lower.contains("services")
405        || lower.contains("daemon")
406        || lower.contains("startup type")
407        || lower.contains("background service")
408        || lower.contains("windows service")
409        || lower.contains("systemctl")
410        || lower.contains("get-service");
411    let asks_processes = lower.contains("process")
412        || lower.contains("processes")
413        || lower.contains("task manager")
414        || lower.contains("what is running")
415        || lower.contains("what's running")
416        || lower.contains("using my ram")
417        || lower.contains("using ram")
418        || lower.contains("using my cpu")
419        || lower.contains("top memory")
420        || lower.contains("top ram")
421        || lower.contains("high memory")
422        || lower.contains("resource-heavy processes")
423        || lower.contains("heavy hitters")
424        || (lower.contains("using the most") && (lower.contains("cpu") || lower.contains("ram") || lower.contains("memory")))
425        || (lower.contains("most cpu") || lower.contains("most ram") || lower.contains("most memory"))
426        || (lower.contains("hitting") && (lower.contains("cpu") || lower.contains("ram") || lower.contains("disk")));
427    let asks_toolchains = lower.contains("developer tools")
428        || lower.contains("toolchains")
429        || (lower.contains("installed") && lower.contains("version"))
430        || (lower.contains("detect") && lower.contains("version"));
431    let asks_ports = lower.contains("listening on port")
432        || lower.contains("listening port")
433        || lower.contains("open port")
434        || lower.contains("port 3000")
435        || lower.contains("port ")
436        || lower.contains("listening on ")
437        || lower.contains("exposed")
438        || lower.contains("what is listening")
439        || (lower.contains("listening") && lower.contains("port"));
440    let asks_repo_doctor = lower.contains("repo doctor")
441        || lower.contains("repository doctor")
442        || lower.contains("workspace health")
443        || lower.contains("repo health")
444        || lower.contains("workspace sanity")
445        || (lower.contains("git state")
446            && (lower.contains("release artifacts")
447                || lower.contains("build markers")
448                || lower.contains("hematite memory")));
449    let asks_directory = lower.contains("directory")
450        || lower.contains("folder")
451        || lower.contains("how big")
452        || lower.contains("biggest");
453    let asks_broad_readiness = lower.contains("local development")
454        || lower.contains("ready for local development")
455        || (lower.contains("machine") && lower.contains("ready"))
456        || (lower.contains("computer") && lower.contains("ready"));
457    let asks_os_config = lower.contains("firewall")
458        || lower.contains("power plan")
459        || lower.contains("power settings")
460        || lower.contains("powercfg")
461        || lower.contains("uptime")
462        || lower.contains("reboot")
463        || lower.contains("boot time")
464        || lower.contains("last boot");
465    let asks_health_report = lower.contains("health report")
466        || lower.contains("system health")
467        || (lower.contains("how") && lower.contains("machine") && lower.contains("doing"))
468        || (lower.contains("status") && lower.contains("report") && !lower.contains("git"));
469    let asks_updates = lower.contains("up to date")
470        || lower.contains("windows update")
471        || lower.contains("pending update")
472        || lower.contains("update available")
473        || lower.contains("check for update")
474        || lower.contains("latest update")
475        || (lower.contains("update")
476            && (lower.contains("my pc")
477                || lower.contains("my computer")
478                || lower.contains("my machine")));
479    let asks_security = lower.contains("antivirus")
480        || lower.contains("defender")
481        || lower.contains("virus protection")
482        || lower.contains("malware")
483        || lower.contains("windows security")
484        || lower.contains("uac")
485        || lower.contains("windows activated")
486        || lower.contains("activation status")
487        || (lower.contains("protected") && (lower.contains("pc") || lower.contains("computer")))
488        || (lower.contains("security")
489            && !lower.contains("git")
490            && !lower.contains("ssh")
491            && !lower.contains("token"));
492    let asks_pending_reboot = lower.contains("need to restart")
493        || lower.contains("need to reboot")
494        || lower.contains("requires restart")
495        || lower.contains("requires a reboot")
496        || lower.contains("reboot required")
497        || lower.contains("restart required")
498        || lower.contains("pending restart")
499        || lower.contains("pending reboot")
500        || (lower.contains("restart")
501            && (lower.contains("waiting")
502                || lower.contains("queued")
503                || lower.contains("required")))
504        || (lower.contains("reboot") && lower.contains("required"))
505        || (lower.contains("reboot") && lower.contains("pending"))
506        || (lower.contains("restart") && lower.contains("pending"));
507    let asks_disk_health = lower.contains("disk health")
508        || lower.contains("drive health")
509        || lower.contains("hard drive dying")
510        || lower.contains("smart status")
511        || lower.contains("drive failing")
512        || lower.contains("drive fail")
513        || (lower.contains("dying") && (lower.contains("drive") || lower.contains("disk")))
514        || (lower.contains("healthy")
515            && (lower.contains("drive")
516                || lower.contains("disk")
517                || lower.contains("ssd")
518                || lower.contains("hdd")));
519    let asks_battery = lower.contains("battery")
520        || lower.contains("battery life")
521        || lower.contains("battery health")
522        || lower.contains("battery wear")
523        || lower.contains("charge level")
524        || lower.contains("how long until")
525        || (lower.contains("dying") && lower.contains("batter"));
526    let asks_recent_crashes = lower.contains("crash")
527        || lower.contains("bsod")
528        || lower.contains("blue screen")
529        || lower.contains("why did my pc restart")
530        || lower.contains("unexpected restart")
531        || lower.contains("sudden restart")
532        || lower.contains("kernel panic")
533        || lower.contains("app crash")
534        || (lower.contains("restart") && lower.contains("itself"))
535        || (lower.contains("restart") && lower.contains("by itself"));
536    let asks_log_check = lower.contains("event log")
537        || lower.contains("windows log")
538        || lower.contains("system log")
539        || lower.contains("error log")
540        || lower.contains("recent errors")
541        || lower.contains("recent warnings")
542        || lower.contains("recent events")
543        || lower.contains("event viewer")
544        || lower.contains("journald")
545        || lower.contains("journal log")
546        || lower.contains("show me errors")
547        || lower.contains("show me warnings")
548        || (lower.contains("log") && lower.contains("error"))
549        || (lower.contains("log") && lower.contains("warning"))
550        || (lower.contains("what errors") && lower.contains("log"));
551    let asks_scheduled_tasks = lower.contains("scheduled task")
552        || lower.contains("scheduled tasks")
553        || lower.contains("task scheduler")
554        || lower.contains("what runs on a timer")
555        || lower.contains("what runs at")
556        || lower.contains("cron job")
557        || lower.contains("background task");
558    let asks_dev_conflicts = lower.contains("dev conflict")
559        || lower.contains("environment conflict")
560        || lower.contains("toolchain conflict")
561        || lower.contains("version conflict")
562        || lower.contains("path conflict")
563        || lower.contains("duplicate path")
564        || (lower.contains("python") && lower.contains("wrong version"))
565        || (lower.contains("node") && lower.contains("wrong version"))
566        || lower.contains("conda shadow")
567        || lower.contains("dev environment clean");
568    let asks_disk_benchmark = lower.contains("benchmark")
569        || lower.contains("stress test")
570        || lower.contains("load test")
571        || lower.contains("intensity report")
572        || lower.contains("io intensity")
573        || lower.contains("disk intensity")
574        || lower.contains("thrash")
575        || lower.contains("latency report");
576    let asks_storage = lower.contains("storage")
577        || lower.contains("disk space")
578        || lower.contains("drive capacity")
579        || lower.contains("free space")
580        || lower.contains("how much space")
581        || lower.contains("space left")
582        || lower.contains("running out of space")
583        || lower.contains("disk usage")
584        || lower.contains("storage usage")
585        || lower.contains("how much disk")
586        || lower.contains("how full")
587        || lower.contains("cache size")
588        || (lower.contains("drive") && lower.contains("usage"))
589        || (lower.contains("drives") && lower.contains("usage"))
590        || (lower.contains("where") && lower.contains("space") && lower.contains("go"));
591    let asks_resource_load = lower.contains("resource load")
592        || lower.contains("system load")
593        || lower.contains("performance")
594        || lower.contains("utilization")
595        || lower.contains("usage report")
596        || lower.contains("performance report")
597        || lower.contains("what is my load")
598        || lower.contains("current load")
599        || lower.contains("why is it slow")
600        || lower.contains("why is it laggy")
601        || lower.contains("is it working hard")
602        || lower.contains("high cpu")
603        || lower.contains("high ram")
604        || lower.contains("cpu load")
605        || lower.contains("heavy hitters")
606        || (lower.contains("resource") && lower.contains("usage"));
607
608    let asks_connectivity = lower.contains("internet")
609        || lower.contains("online")
610        || lower.contains("connectivity")
611        || lower.contains("am i connected")
612        || lower.contains("ping google")
613        || lower.contains("reach the internet")
614        || lower.contains("internet access")
615        || lower.contains("no internet")
616        || lower.contains("internet down")
617        || lower.starts_with("ping ")
618        || lower.contains(" ping ")
619        || (lower.contains("check") && lower.contains("connection"))
620        || (lower.contains("dns") && (lower.contains("resolv") || lower.contains("working")));
621    let asks_wifi = lower.contains("wi-fi")
622        || lower.contains("wifi")
623        || lower.contains("wireless")
624        || lower.contains("wlan")
625        || lower.contains("signal strength")
626        || lower.contains("ssid")
627        || lower.contains("access point")
628        || (lower.contains("wireless") && lower.contains("connect"));
629    let asks_connections = lower.contains("tcp connection")
630        || lower.contains("active connection")
631        || lower.contains("established connection")
632        || lower.contains("socket")
633        || lower.contains("netstat")
634        || (lower.contains("connection") && lower.contains("active"))
635        || (lower.contains("connection") && lower.contains("open"));
636    let asks_vpn = lower.contains("vpn")
637        || lower.contains("virtual private network")
638        || (lower.contains("tunnel") && (lower.contains("network") || lower.contains("vpn")));
639    let asks_proxy = lower.contains("proxy")
640        || lower.contains("proxy setting")
641        || lower.contains("winhttp proxy")
642        || lower.contains("system proxy")
643        || (lower.contains("routed") && lower.contains("proxy"));
644    let asks_firewall_rules = (lower.contains("firewall")
645        && (lower.contains("rule")
646            || lower.contains("block")
647            || lower.contains("allow")
648            || lower.contains("inbound")
649            || lower.contains("outbound")))
650        || lower.contains("blocked port")
651        || lower.contains("firewall rule");
652    let asks_traceroute = lower.contains("traceroute")
653        || lower.contains("tracert")
654        || lower.contains("tracepath")
655        || lower.contains("trace route")
656        || lower.contains("trace the route")
657        || lower.contains("trace the path")
658        || lower.contains("network path")
659        || lower.contains("how many hops")
660        || lower.contains("where does traffic go")
661        || (lower.contains("trace") && lower.contains("hop"))
662        || (lower.contains("route") && lower.contains("traffic"))
663        || (lower.contains("trace") && lower.contains("8.8.8.8"))
664        || (lower.contains("path") && lower.contains("8.8.8.8"));
665    let asks_dns_cache = lower.contains("dns cache")
666        || lower.contains("cached dns")
667        || lower.contains("dns lookup cache")
668        || lower.contains("displaydns")
669        || lower.contains("/displaydns")
670        || lower.contains("get-dnsclientcache")
671        || lower.contains("dns entries")
672        || (lower.contains("dns") && lower.contains("cached"));
673    let asks_arp = lower.contains("arp -")
674        || lower.contains("arp table")
675        || lower.contains("arp cache")
676        || lower.contains("mac address")
677        || lower.contains("neighbor table")
678        || lower.contains("ip to mac")
679        || lower.contains("ip neigh")
680        || (lower.contains("arp")
681            && (lower.contains("who") || lower.contains("entry") || lower.contains("entries")));
682    let asks_route_table = lower.contains("route print")
683        || lower.contains("route table")
684        || lower.contains("routing table")
685        || lower.contains("get-netroute")
686        || lower.contains("default gateway")
687        || lower.contains("network routes")
688        || lower.contains("ip route")
689        || lower.contains("next hop")
690        || (lower.contains("route")
691            && (lower.contains("table") || lower.contains("entry") || lower.contains("entries")));
692    let asks_env = (lower.contains("environment variable")
693        || lower.contains("env var")
694        || lower.contains("env vars")
695        || lower.contains("show env")
696        || lower.contains("list env"))
697        && !lower.contains("env doctor");
698    let asks_hosts_file = lower.contains("hosts file")
699        || lower.contains("/etc/hosts")
700        || lower.contains("etc/hosts")
701        || lower.contains("hosts entry")
702        || lower.contains("hosts entries")
703        || (lower.contains("hosts")
704            && (lower.contains("redirect")
705                || lower.contains("block")
706                || lower.contains("loopback")));
707    let asks_docker = lower.contains("docker")
708        || lower.contains("container")
709        || lower.contains("docker compose")
710        || lower.contains("docker ps")
711        || lower.contains("running container");
712    let asks_wsl = lower.contains("wsl")
713        || lower.contains("windows subsystem")
714        || lower.contains("linux distro")
715        || lower.contains("ubuntu on windows")
716        || (lower.contains("subsystem") && lower.contains("linux"));
717    let asks_ssh = (lower.contains("ssh") && !lower.contains("ssh key") && !lower.contains("git"))
718        || lower.contains("sshd")
719        || lower.contains("ssh config")
720        || lower.contains("ssh server")
721        || lower.contains("ssh client")
722        || lower.contains("known_hosts")
723        || lower.contains("authorized_keys")
724        || lower.contains("ssh key")
725        || (lower.contains("ssh")
726            && (lower.contains("running")
727                || lower.contains("service")
728                || lower.contains("port 22")));
729    let asks_installed_software = lower.contains("installed software")
730        || lower.contains("installed program")
731        || lower.contains("installed app")
732        || lower.contains("installed package")
733        || lower.contains("what is installed")
734        || lower.contains("what's installed")
735        || lower.contains("winget list")
736        || lower.contains("list programs")
737        || (lower.contains("installed")
738            && (lower.contains("on this machine")
739                || lower.contains("on my machine")
740                || lower.contains("on my pc")));
741    let asks_databases = lower.contains("postgres")
742        || lower.contains("postgresql")
743        || lower.contains("mysql")
744        || lower.contains("mariadb")
745        || lower.contains("mongodb")
746        || lower.contains("mongo")
747        || lower.contains("redis")
748        || lower.contains("sql server")
749        || lower.contains("mssql")
750        || lower.contains("sqlite")
751        || lower.contains("elasticsearch")
752        || lower.contains("cassandra")
753        || lower.contains("couchdb")
754        || (lower.contains("database")
755            && (lower.contains("running")
756                || lower.contains("service")
757                || lower.contains("installed")
758                || lower.contains("up")
759                || lower.contains("local")))
760        || lower.contains("db service")
761        || lower.contains("database server")
762        || (lower.contains("is")
763            && lower.contains("running")
764            && (lower.contains("db") || lower.contains("database")));
765    let asks_git_config = (lower.contains("git config")
766        || lower.contains("git configuration")
767        || lower.contains("git global")
768        || (lower.contains("git") && lower.contains("user.name"))
769        || (lower.contains("git") && lower.contains("user.email"))
770        || (lower.contains("git") && lower.contains("signing"))
771        || (lower.contains("git") && lower.contains("credential"))
772        || lower.contains("git aliases"))
773        && !lower.contains("github");
774    let asks_user_accounts = lower.contains("local user")
775        || lower.contains("user account")
776        || lower.contains("who is logged in")
777        || lower.contains("who is logged on")
778        || lower.contains("logged in user")
779        || lower.contains("logged on user")
780        || lower.contains("admin group")
781        || lower.contains("administrators group")
782        || lower.contains("local admin")
783        || lower.contains("who has admin")
784        || lower.contains("running as admin")
785        || lower.contains("is this elevated")
786        || lower.contains("active sessions")
787        || lower.contains("logon session")
788        || lower.contains("net user")
789        || lower.contains("get-localuser");
790    let asks_audit_policy = lower.contains("audit policy")
791        || lower.contains("auditpol")
792        || lower.contains("audit log")
793        || lower.contains("what is being logged")
794        || lower.contains("security audit")
795        || lower.contains("logon event")
796        || lower.contains("audit category")
797        || lower.contains("event auditing");
798    let asks_shares = lower.contains("smb share")
799        || lower.contains("network share")
800        || lower.contains("shared folder")
801        || lower.contains("mapped drive")
802        || lower.contains("mapped network drive")
803        || lower.contains("get-smbshare")
804        || lower.contains("what is shared")
805        || lower.contains("what am i sharing")
806        || lower.contains("smb1")
807        || lower.contains("smb signing")
808        || lower.contains("nfs export");
809    let asks_dns_servers = (lower.contains("dns server")
810        || lower.contains("dns resolver")
811        || lower.contains("nameserver")
812        || lower.contains("which dns")
813        || lower.contains("what dns")
814        || lower.contains("dns over https")
815        || lower.contains("doh")
816        || lower.contains("dns search suffix")
817        || lower.contains("configured dns")
818        || lower.contains("get-dnsclientserveraddress"))
819        && !lower.contains("dns cache");
820    let asks_bitlocker = lower.contains("bitlocker")
821        || (lower.contains("drive") && lower.contains("encrypt"))
822        || (lower.contains("disk") && lower.contains("encrypt"))
823        || lower.contains("encryption status");
824    let asks_rdp = lower.contains("rdp")
825        || lower.contains("remote desktop")
826        || (lower.contains("remote") && lower.contains("access") && !lower.contains("git"));
827    let asks_shadow_copies = lower.contains("shadow copy")
828        || lower.contains("shadow copies")
829        || lower.contains("vss")
830        || lower.contains("snapshot")
831        || lower.contains("restore point");
832    let asks_pagefile = lower.contains("pagefile")
833        || lower.contains("page file")
834        || lower.contains("virtual memory")
835        || lower.contains("swap file")
836        || (lower.contains("paging") && lower.contains("file"));
837    let asks_windows_features = (lower.contains("window") && lower.contains("feature"))
838        || lower.contains("optional feature")
839        || lower.contains("iis")
840        || lower.contains("hyper-v")
841        || (lower.contains("feature")
842            && (lower.contains("install")
843                || lower.contains("enabled")
844                || lower.contains("turn on")));
845    let asks_printers =
846        lower.contains("printer") || lower.contains("print queue") || lower.contains("get-printer");
847    let asks_winrm = lower.contains("winrm")
848        || lower.contains("psremoting")
849        || (lower.contains("ps") && lower.contains("remoting"))
850        || (lower.contains("remote") && lower.contains("management") && !lower.contains("rdp"));
851    let asks_network_stats = (lower.contains("network") && lower.contains("stat"))
852        || (lower.contains("adapter") && lower.contains("stat"))
853        || (lower.contains("nic") && lower.contains("stat"))
854        || lower.contains("throughput")
855        || lower.contains("packet loss")
856        || lower.contains("dropped packet");
857    let asks_udp_ports = lower.contains("udp port")
858        || lower.contains("udp listener")
859        || (lower.contains("udp")
860            && (lower.contains("port") || lower.contains("listen") || lower.contains("open")));
861
862    if asks_disk_benchmark {
863        Some("disk_benchmark")
864    } else if asks_storage {
865        Some("storage")
866    } else if asks_fix_plan {
867        Some("fix_plan")
868    } else if asks_gpo {
869        Some("gpo")
870    } else if asks_certificates {
871        Some("certificates")
872    } else if asks_integrity {
873        Some("integrity")
874    } else if asks_domain {
875        Some("domain")
876    } else if asks_device_health {
877        Some("device_health")
878    } else if asks_drivers {
879        Some("drivers")
880    } else if asks_peripherals {
881        Some("peripherals")
882    } else if asks_user_accounts {
883        Some("user_accounts")
884    } else if asks_sessions {
885        Some("sessions")
886    } else if asks_virtualization {
887        Some("hardware")
888    } else if asks_startup {
889        Some("startup_items")
890    } else if asks_bitlocker {
891        Some("bitlocker")
892    } else if asks_rdp {
893        Some("rdp")
894    } else if asks_shadow_copies {
895        Some("shadow_copies")
896    } else if asks_pagefile {
897        Some("pagefile")
898    } else if asks_windows_features {
899        Some("windows_features")
900    } else if asks_printers {
901        Some("printers")
902    } else if asks_winrm {
903        Some("winrm")
904    } else if asks_network_stats {
905        Some("network_stats")
906    } else if asks_udp_ports {
907        Some("udp_ports")
908    } else if asks_updates {
909        Some("updates")
910    } else if asks_audit_policy {
911        Some("audit_policy")
912    } else if asks_security {
913        Some("security")
914    } else if asks_pending_reboot {
915        Some("pending_reboot")
916    } else if asks_disk_health {
917        Some("disk_health")
918    } else if asks_battery {
919        Some("battery")
920    } else if asks_recent_crashes {
921        Some("recent_crashes")
922    } else if asks_log_check {
923        Some("log_check")
924    } else if asks_scheduled_tasks {
925        Some("scheduled_tasks")
926    } else if asks_dev_conflicts {
927        Some("dev_conflicts")
928    } else if (asks_path && asks_toolchains)
929        || (mentions_host_inspection_question(&lower) && asks_broad_readiness)
930    {
931        Some("summary")
932    } else if asks_env_doctor {
933        Some("env_doctor")
934    } else if asks_dns_servers {
935        Some("dns_servers")
936    } else if asks_connectivity {
937        Some("connectivity")
938    } else if asks_wifi {
939        Some("wifi")
940    } else if asks_connections {
941        Some("connections")
942    } else if asks_vpn {
943        Some("vpn")
944    } else if asks_proxy {
945        Some("proxy")
946    } else if asks_firewall_rules {
947        Some("firewall_rules")
948    } else if asks_traceroute {
949        Some("traceroute")
950    } else if asks_dns_cache {
951        Some("dns_cache")
952    } else if asks_arp {
953        Some("arp")
954    } else if asks_route_table {
955        Some("route_table")
956    } else if asks_databases {
957        Some("databases")
958    } else if asks_docker {
959        Some("docker")
960    } else if asks_wsl {
961        Some("wsl")
962    } else if asks_ssh {
963        Some("ssh")
964    } else if asks_git_config {
965        Some("git_config")
966    } else if asks_installed_software {
967        Some("installed_software")
968    } else if asks_env {
969        Some("env")
970    } else if asks_shares {
971        Some("shares")
972    } else if asks_hosts_file {
973        Some("hosts_file")
974    } else if asks_network {
975        Some("network")
976    } else if asks_services {
977        Some("services")
978    } else if asks_ports {
979        Some("ports")
980    } else if asks_processes {
981        Some("processes")
982    } else if asks_repo_doctor {
983        Some("repo_doctor")
984    } else if lower.contains("desktop") {
985        Some("desktop")
986    } else if lower.contains("downloads") {
987        Some("downloads")
988    } else if asks_path {
989        Some("path")
990    } else if asks_toolchains {
991        Some("toolchains")
992    } else if asks_os_config {
993        Some("os_config")
994    } else if asks_resource_load {
995        Some("resource_load")
996    } else if asks_health_report {
997        Some("health_report")
998    } else if asks_directory {
999        Some("directory")
1000    } else if mentions_host_inspection_question(&lower) {
1001        Some("summary")
1002    } else {
1003        None
1004    }
1005}
1006
1007/// Returns all distinct inspect_host topics detected in a user prompt.
1008/// Used by the harness to pre-run multiple topics when the user asks for several at once,
1009/// so the model receives all results and only needs to synthesize rather than orchestrate.
1010pub fn all_host_inspection_topics(user_input: &str) -> Vec<&'static str> {
1011    // All topic detectors in priority order — ordered so more specific topics come
1012    // before generic fallbacks (e.g. traceroute before network).
1013    let detectors: &[(&str, fn(&str) -> bool)] = &[
1014        ("fix_plan", |l| {
1015            l.contains("fix")
1016                && (l.contains("cargo")
1017                    || l.contains("port ")
1018                    || l.contains("lm studio")
1019                    || l.contains("toolchain"))
1020        }),
1021        ("updates", |l| {
1022            l.contains("up to date")
1023                || l.contains("windows update")
1024                || l.contains("pending update")
1025                || l.contains("update available")
1026        }),
1027        ("security", |l| {
1028            l.contains("antivirus")
1029                || l.contains("defender")
1030                || l.contains("uac")
1031                || (l.contains("security") && !l.contains("git") && !l.contains("ssh"))
1032        }),
1033        ("pending_reboot", |l| {
1034            l.contains("pending reboot")
1035                || l.contains("pending restart")
1036                || l.contains("need to restart")
1037                || l.contains("reboot required")
1038                || (l.contains("reboot") && l.contains("pending"))
1039                || (l.contains("restart") && l.contains("pending"))
1040        }),
1041        ("disk_health", |l| {
1042            l.contains("disk health")
1043                || l.contains("drive health")
1044                || l.contains("smart status")
1045                || (l.contains("healthy")
1046                    && (l.contains("drive") || l.contains("disk") || l.contains("ssd")))
1047        }),
1048        ("battery", |l| l.contains("battery")),
1049        ("recent_crashes", |l| {
1050            l.contains("crash") || l.contains("bsod") || l.contains("blue screen")
1051        }),
1052        ("scheduled_tasks", |l| {
1053            l.contains("scheduled task") || l.contains("task scheduler")
1054        }),
1055        ("dev_conflicts", |l| {
1056            l.contains("dev conflict")
1057                || l.contains("toolchain conflict")
1058                || l.contains("duplicate path")
1059        }),
1060        ("storage", |l| {
1061            l.contains("disk space")
1062                || l.contains("storage")
1063                || l.contains("drive capacity")
1064                || l.contains("cache size")
1065        }),
1066        ("hardware", |l| {
1067            l.contains("cpu model")
1068                || l.contains("ram size")
1069                || l.contains("hardware spec")
1070                || (l.contains("what hardware") && l.contains("have"))
1071        }),
1072        ("health_report", |l| {
1073            l.contains("health report") || l.contains("system health")
1074        }),
1075        ("resource_load", |l| {
1076            l.contains("resource load")
1077                || l.contains("cpu load")
1078                || l.contains("ram %")
1079                || l.contains("cpu %")
1080                || l.contains("performance")
1081        }),
1082        ("processes", |l| {
1083            l.contains("process")
1084                || l.contains("task manager")
1085                || l.contains("what is running")
1086                || l.contains("using my ram")
1087                || l.contains("hitting the disk")
1088                || l.contains("disk thrasher")
1089        }),
1090        ("services", |l| {
1091            l.contains("service") || l.contains("daemon") || l.contains("windows service")
1092        }),
1093        ("ports", |l| {
1094            l.contains("listening port")
1095                || l.contains("open port")
1096                || l.contains("what is on port")
1097                || l.contains("port 3000")
1098                || (l.contains("listening") && l.contains("port"))
1099        }),
1100        ("traceroute", |l| {
1101            l.contains("traceroute")
1102                || l.contains("tracert")
1103                || l.contains("trace route")
1104                || l.contains("trace the path")
1105                || l.contains("network path")
1106                || l.contains("how many hops")
1107                || (l.contains("trace") && l.contains("hop"))
1108        }),
1109        ("dns_cache", |l| {
1110            l.contains("dns cache")
1111                || l.contains("cached dns")
1112                || l.contains("displaydns")
1113                || (l.contains("dns") && l.contains("cached"))
1114        }),
1115        ("arp", |l| {
1116            l.contains("arp table")
1117                || l.contains("arp cache")
1118                || l.contains("mac address")
1119                || l.contains("ip to mac")
1120                || l.contains("arp -")
1121        }),
1122        ("route_table", |l| {
1123            l.contains("route table")
1124                || l.contains("routing table")
1125                || l.contains("route print")
1126                || l.contains("network route")
1127                || l.contains("next hop")
1128        }),
1129        ("connectivity", |l| {
1130            l.contains("internet")
1131                || l.contains("am i connected")
1132                || l.contains("ping google")
1133                || l.contains("internet access")
1134                || l.contains("no internet")
1135        }),
1136        ("wifi", |l| {
1137            l.contains("wi-fi")
1138                || l.contains("wifi")
1139                || l.contains("wireless")
1140                || l.contains("ssid")
1141                || l.contains("signal strength")
1142        }),
1143        ("connections", |l| {
1144            l.contains("tcp connection")
1145                || l.contains("active connection")
1146                || l.contains("netstat")
1147                || l.contains("open socket")
1148                || (l.contains("established") && l.contains("connection"))
1149        }),
1150        ("vpn", |l| {
1151            l.contains("vpn") || l.contains("virtual private network")
1152        }),
1153        ("proxy", |l| {
1154            l.contains("proxy setting") || l.contains("system proxy") || l.contains("winhttp proxy")
1155        }),
1156        ("firewall_rules", |l| {
1157            (l.contains("firewall")
1158                && (l.contains("rule") || l.contains("inbound") || l.contains("outbound")))
1159                || l.contains("firewall rule")
1160        }),
1161        ("network", |l| {
1162            l.contains("network adapter")
1163                || l.contains("ip address")
1164                || l.contains("ipconfig")
1165                || l.contains("gateway")
1166                || l.contains("subnet")
1167        }),
1168        ("env_doctor", |l| {
1169            l.contains("env doctor")
1170                || l.contains("environment doctor")
1171                || l.contains("package manager")
1172                || l.contains("path drift")
1173        }),
1174        ("os_config", |l| {
1175            l.contains("power plan")
1176                || l.contains("uptime")
1177                || l.contains("boot time")
1178                || l.contains("last boot")
1179        }),
1180        ("path", |l| {
1181            l.contains("path entries") || l.contains("raw path")
1182        }),
1183        ("toolchains", |l| {
1184            l.contains("developer tools")
1185                || l.contains("toolchains")
1186                || (l.contains("installed") && l.contains("version"))
1187        }),
1188        ("docker", |l| {
1189            l.contains("docker") || l.contains("container") || l.contains("running container")
1190        }),
1191        ("wsl", |l| {
1192            l.contains("wsl")
1193                || l.contains("windows subsystem")
1194                || (l.contains("subsystem") && l.contains("linux"))
1195        }),
1196        ("ssh", |l| {
1197            l.contains("ssh")
1198                || l.contains("sshd")
1199                || l.contains("known_hosts")
1200                || l.contains("authorized_keys")
1201        }),
1202        ("git_config", |l| {
1203            (l.contains("git config") || l.contains("git global") || l.contains("git aliases"))
1204                && !l.contains("github")
1205        }),
1206        ("installed_software", |l| {
1207            l.contains("installed software")
1208                || l.contains("installed program")
1209                || l.contains("what is installed")
1210                || l.contains("what's installed")
1211                || l.contains("winget list")
1212        }),
1213        ("env", |l| {
1214            (l.contains("environment variable") || l.contains("env var") || l.contains("env vars"))
1215                && !l.contains("env doctor")
1216        }),
1217        ("hosts_file", |l| {
1218            l.contains("hosts file") || l.contains("/etc/hosts") || l.contains("hosts entry")
1219        }),
1220        ("databases", |l| {
1221            l.contains("postgres")
1222                || l.contains("mysql")
1223                || l.contains("mariadb")
1224                || l.contains("mongodb")
1225                || l.contains("redis")
1226                || l.contains("sqlite")
1227                || l.contains("sql server")
1228                || l.contains("elasticsearch")
1229                || (l.contains("database") && (l.contains("running") || l.contains("service")))
1230        }),
1231        ("user_accounts", |l| {
1232            l.contains("local user")
1233                || l.contains("user account")
1234                || l.contains("who is logged")
1235                || l.contains("admin group")
1236                || l.contains("local admin")
1237                || l.contains("active sessions")
1238                || l.contains("running as admin")
1239        }),
1240        ("audit_policy", |l| {
1241            l.contains("audit policy")
1242                || l.contains("auditpol")
1243                || l.contains("what is being logged")
1244                || l.contains("security audit")
1245                || l.contains("event auditing")
1246        }),
1247        ("shares", |l| {
1248            l.contains("smb share")
1249                || l.contains("network share")
1250                || l.contains("shared folder")
1251                || l.contains("mapped drive")
1252                || l.contains("smb1")
1253                || l.contains("nfs export")
1254        }),
1255        ("dns_servers", |l| {
1256            (l.contains("dns server")
1257                || l.contains("dns resolver")
1258                || l.contains("nameserver")
1259                || l.contains("which dns")
1260                || l.contains("dns over https")
1261                || l.contains("configured dns"))
1262                && !l.contains("dns cache")
1263        }),
1264        ("bitlocker", |l| {
1265            l.contains("bitlocker")
1266                || (l.contains("drive") && l.contains("encrypt"))
1267                || (l.contains("disk") && l.contains("encrypt"))
1268                || l.contains("encryption status")
1269        }),
1270        ("rdp", |l| {
1271            l.contains("rdp")
1272                || l.contains("remote desktop")
1273                || (l.contains("remote") && l.contains("access") && !l.contains("git"))
1274        }),
1275        ("shadow_copies", |l| {
1276            l.contains("shadow copy")
1277                || l.contains("shadow copies")
1278                || l.contains("vss")
1279                || l.contains("snapshot")
1280                || l.contains("restore point")
1281        }),
1282        ("pagefile", |l| {
1283            l.contains("pagefile")
1284                || l.contains("page file")
1285                || l.contains("virtual memory")
1286                || l.contains("swap file")
1287        }),
1288        ("windows_features", |l| {
1289            (l.contains("window") && l.contains("feature"))
1290                || l.contains("optional feature")
1291                || l.contains("iis")
1292                || l.contains("hyper-v")
1293                || (l.contains("feature") && (l.contains("install") || l.contains("enabled")))
1294        }),
1295        ("printers", |l| {
1296            l.contains("printer") || l.contains("print queue") || l.contains("get-printer")
1297        }),
1298        ("winrm", |l| {
1299            l.contains("winrm")
1300                || l.contains("psremoting")
1301                || (l.contains("remote") && l.contains("management") && !l.contains("rdp"))
1302        }),
1303        ("network_stats", |l| {
1304            (l.contains("network") && l.contains("stat"))
1305                || (l.contains("adapter") && l.contains("stat"))
1306                || l.contains("throughput")
1307                || l.contains("packet loss")
1308                || l.contains("dropped packet")
1309        }),
1310        ("startup_items", |l| {
1311            l.contains("startup") || l.contains("boot program") || l.contains("autorun")
1312        }),
1313        ("udp_ports", |l| {
1314            l.contains("udp port")
1315                || l.contains("udp listener")
1316                || (l.contains("udp") && l.contains("listening"))
1317        }),
1318        ("gpo", |l| {
1319            l.contains("gpo") || l.contains("group policy") || l.contains("gpresult")
1320        }),
1321        ("certificates", |l| {
1322            l.contains("cert") || l.contains("ssl") || l.contains("thumbprint")
1323        }),
1324        ("integrity", |l| {
1325            l.contains("integrity") || l.contains("sfc") || l.contains("dism")
1326        }),
1327        ("domain", |l| {
1328            l.contains("domain")
1329                || l.contains("workgroup")
1330                || l.contains("netbios")
1331                || l.contains("active directory")
1332        }),
1333        ("device_health", |l| {
1334            l.contains("device health")
1335                || l.contains("hardware error")
1336                || l.contains("yellow bang")
1337                || l.contains("malfunctioning")
1338        }),
1339        ("drivers", |l| {
1340            l.contains("driver") || l.contains("system driver")
1341        }),
1342        ("peripherals", |l| {
1343            l.contains("peripheral")
1344                || l.contains("usb")
1345                || l.contains("keyboard")
1346                || l.contains("mouse")
1347                || l.contains("monitor")
1348        }),
1349        ("sessions", |l| {
1350            l.contains("session") || l.contains("who is logged") || l.contains("active login")
1351        }),
1352        ("hardware", |l| {
1353            l.contains("virtualization")
1354                || l.contains("hypervisor")
1355                || l.contains("vt-x")
1356                || l.contains("slat")
1357        }),
1358    ];
1359
1360    let lower = user_input.to_lowercase();
1361    let mut topics: Vec<&'static str> = Vec::new();
1362    for (topic, check) in detectors {
1363        if check(&lower) && !topics.contains(topic) {
1364            topics.push(topic);
1365        }
1366    }
1367    topics
1368}
1369
1370pub(crate) fn preferred_maintainer_workflow(user_input: &str) -> Option<&'static str> {
1371    let lower = user_input.to_ascii_lowercase();
1372    let asks_cleanup = contains_any(
1373        &lower,
1374        &[
1375            "run my cleanup",
1376            "run the cleanup",
1377            "run cleanup",
1378            "deep clean",
1379            "prune dist",
1380            "clean.ps1",
1381            "cleanup script",
1382            "cleanup workflow",
1383            "clean up scripts",
1384        ],
1385    );
1386    let asks_package = contains_any(
1387        &lower,
1388        &[
1389            "rebuild local portable",
1390            "rebuild the portable",
1391            "run the local build",
1392            "run the portable",
1393            "package-windows.ps1",
1394            "package windows",
1395            "build installer",
1396            "overwrite the portable",
1397            "refresh the portable",
1398            "update path",
1399            "update path with the portable",
1400        ],
1401    );
1402    let asks_release = contains_any(
1403        &lower,
1404        &[
1405            "run the release flow",
1406            "regular workflow",
1407            "cut the release",
1408            "ship it",
1409            "release.ps1",
1410            "bump to ",
1411            "tag it",
1412            "full tag and everything",
1413            "publish crates",
1414        ],
1415    );
1416
1417    if asks_cleanup {
1418        Some("clean")
1419    } else if asks_package {
1420        Some("package_windows")
1421    } else if asks_release {
1422        Some("release")
1423    } else {
1424        None
1425    }
1426}
1427
1428pub(crate) fn preferred_workspace_workflow(user_input: &str) -> Option<&'static str> {
1429    let lower = user_input.to_ascii_lowercase();
1430    let asks_project_scope = contains_any(
1431        &lower,
1432        &[
1433            "this repo",
1434            "this repository",
1435            "this project",
1436            "current project",
1437            "current repo",
1438            "workspace",
1439            "in this folder",
1440            "here",
1441        ],
1442    );
1443    let asks_build = contains_any(
1444        &lower,
1445        &[
1446            "run the build",
1447            "build this project",
1448            "build this repo",
1449            "run build",
1450            "compile this project",
1451            "cargo build",
1452            "npm run build",
1453            "pnpm run build",
1454            "yarn build",
1455            "go build",
1456            "gradlew build",
1457        ],
1458    );
1459    let asks_test = contains_any(
1460        &lower,
1461        &[
1462            "run the tests",
1463            "run tests",
1464            "test this project",
1465            "test this repo",
1466            "run the test suite",
1467            "cargo test",
1468            "npm test",
1469            "pnpm test",
1470            "yarn test",
1471            "pytest",
1472            "go test",
1473            "gradlew test",
1474        ],
1475    );
1476    let asks_lint = contains_any(
1477        &lower,
1478        &[
1479            "run lint",
1480            "lint this project",
1481            "lint this repo",
1482            "cargo clippy",
1483            "npm run lint",
1484            "pnpm run lint",
1485            "yarn lint",
1486        ],
1487    );
1488    let asks_fix = contains_any(
1489        &lower,
1490        &[
1491            "run fix",
1492            "fix formatting",
1493            "run formatter",
1494            "cargo fmt",
1495            "npm run fix",
1496            "pnpm run fix",
1497            "yarn fix",
1498        ],
1499    );
1500    let asks_script = contains_any(
1501        &lower,
1502        &[
1503            "npm run ",
1504            "pnpm run ",
1505            "yarn ",
1506            "bun run ",
1507            "make ",
1508            "just ",
1509            "task ",
1510            "scripts/",
1511            ".\\scripts\\",
1512            "./scripts/",
1513            ".ps1",
1514            ".sh",
1515            ".py",
1516            ".cmd",
1517            ".bat",
1518        ],
1519    );
1520
1521    if asks_build
1522        && (asks_project_scope
1523            || !contains_any(&lower, &["release.ps1", "package-windows.ps1", "clean.ps1"]))
1524    {
1525        Some("build")
1526    } else if asks_test && asks_project_scope {
1527        Some("test")
1528    } else if asks_lint && asks_project_scope {
1529        Some("lint")
1530    } else if asks_fix && asks_project_scope {
1531        Some("fix")
1532    } else if asks_script && !preferred_maintainer_workflow(user_input).is_some() {
1533        Some("script")
1534    } else if (asks_test || asks_lint || asks_fix)
1535        && !preferred_maintainer_workflow(user_input).is_some()
1536    {
1537        Some(if asks_test {
1538            "test"
1539        } else if asks_lint {
1540            "lint"
1541        } else {
1542            "fix"
1543        })
1544    } else {
1545        None
1546    }
1547}
1548
1549pub(crate) fn looks_like_mutation_request(user_input: &str) -> bool {
1550    let lower = user_input.to_lowercase();
1551    [
1552        "fix ",
1553        "change ",
1554        "edit ",
1555        "modify ",
1556        "update ",
1557        "rename ",
1558        "refactor ",
1559        "patch ",
1560        "rewrite ",
1561        "implement ",
1562        "create a file",
1563        "create file",
1564        "add a file",
1565        "delete ",
1566        "remove ",
1567        "make the change",
1568    ]
1569    .iter()
1570    .any(|needle| lower.contains(needle))
1571}
1572
1573pub(crate) fn classify_query_intent(workflow_mode: WorkflowMode, user_input: &str) -> QueryIntent {
1574    let lower = user_input.to_lowercase();
1575    let trimmed = user_input.trim().to_ascii_lowercase();
1576
1577    let mentions_runtime_trace = contains_any(
1578        &lower,
1579        &[
1580            "trace",
1581            "how does",
1582            "what are the main runtime subsystems",
1583            "how does a user message move",
1584            "separate normal assistant output",
1585            "session reset behavior",
1586            "file references",
1587            "event types",
1588            "channels",
1589        ],
1590    );
1591    let anti_guess = contains_any(&lower, &["do not guess", "if you are unsure"]);
1592    let capability_mode = mentions_capability_question(&lower);
1593    let capability_needs_repo =
1594        capability_mode && capability_question_requires_repo_inspection(&lower);
1595    let host_inspection_mode = preferred_host_inspection_topic(&lower).is_some();
1596    let maintainer_workflow_mode = preferred_maintainer_workflow(&lower).is_some();
1597    let workspace_workflow_mode =
1598        preferred_workspace_workflow(&lower).is_some() && !maintainer_workflow_mode;
1599    let toolchain_mode = contains_any(
1600        &lower,
1601        &[
1602            "tooling discipline",
1603            "best read-only toolchain",
1604            "identify the best tools you actually have",
1605            "concrete read-only investigation plan",
1606            "do not execute the plan",
1607            "available repo-inspection tools",
1608            "tool choice discipline",
1609            "what tools would you choose first",
1610        ],
1611    ) || (lower.contains("which tools") && lower.contains("why"))
1612        || (lower.contains("when would you choose") && lower.contains("tool"));
1613    let architecture_overview_mode = {
1614        let architecture_signals = contains_any(
1615            &lower,
1616            &[
1617                "architecture overview",
1618                "architecture walkthrough",
1619                "full architecture",
1620                "runtime walkthrough",
1621                "control flow",
1622                "tool routing",
1623                "workflow modes",
1624                "repo map behavior",
1625                "mcp policy",
1626                "prompt budgeting",
1627                "compaction",
1628                "file ownership",
1629                "owner file",
1630                "project structure",
1631                "repository structure",
1632            ],
1633        );
1634        let broad = contains_any(
1635            &lower,
1636            &[
1637                "full detailed",
1638                "all in one answer",
1639                "concrete file ownership",
1640                "walk me through",
1641                "major runtime pieces",
1642                "which files own",
1643                "how",
1644                "explain",
1645                "overview",
1646            ],
1647        );
1648        (architecture_signals && broad)
1649            || (lower.contains("runtime")
1650                && lower.contains("workflow")
1651                && (lower.contains("architecture") || lower.contains("tool routing")))
1652            || mentions_broad_system_walkthrough(&lower)
1653    };
1654
1655    let direct_answer = if trimmed == "/about" || mentions_creator_question(&lower) {
1656        Some(DirectAnswerKind::About)
1657    } else if matches!(
1658        trimmed.as_str(),
1659        "who are you" | "who are you?" | "what are you" | "what are you?"
1660    ) || (lower.contains("what is hematite") && !lower.contains("lm studio"))
1661    {
1662        Some(DirectAnswerKind::Identity)
1663    } else if (mentions_stable_product_surface(&lower) || mentions_product_truth_routing(&lower))
1664        && contains_any(
1665            &lower,
1666            &[
1667                "how hematite answers",
1668                "how does hematite answer",
1669                "how hematite handles",
1670                "how does hematite handle",
1671                "how hematite decides",
1672                "how does hematite decide",
1673                "decides whether",
1674                "decide whether",
1675            ],
1676        )
1677    {
1678        Some(DirectAnswerKind::ProductSurface)
1679    } else if mentions_reset_commands(&lower)
1680        && contains_any(
1681            &lower,
1682            &[
1683                "exact difference",
1684                "difference between",
1685                "explain the exact difference",
1686                "what is the difference",
1687            ],
1688        )
1689    {
1690        Some(DirectAnswerKind::SessionResetSemantics)
1691    } else if (lower.contains("reasoning output") || lower.contains("reasoning"))
1692        && contains_any(
1693            &lower,
1694            &["visible chat output", "visible chat", "chat output"],
1695        )
1696    {
1697        Some(DirectAnswerKind::ReasoningSplit)
1698    } else if lower.contains("/ask")
1699        && lower.contains("/code")
1700        && lower.contains("/architect")
1701        && lower.contains("/read-only")
1702        && lower.contains("/auto")
1703        && contains_any(&lower, &["difference", "differences", "what are"])
1704    {
1705        Some(DirectAnswerKind::WorkflowModes)
1706    } else if lower.contains(".hematite/settings.json")
1707        && lower.contains("gemma_native_auto")
1708        && lower.contains("gemma_native_formatting")
1709    {
1710        Some(DirectAnswerKind::GemmaNativeSettings)
1711    } else if contains_any(
1712        &lower,
1713        &[
1714            "skip verification",
1715            "skip build verification",
1716            "commit it immediately",
1717            "commit immediately",
1718        ],
1719    ) && contains_any(
1720        &lower,
1721        &[
1722            "make a code change",
1723            "make the change",
1724            "change the code",
1725            "edit the code",
1726            "edit a file",
1727            "implement",
1728        ],
1729    ) {
1730        Some(DirectAnswerKind::UnsafeWorkflowPressure)
1731    } else if contains_any(&lower, &["/gemma-native", "gemma native"])
1732        && contains_any(&lower, &["what does", "what is", "how does", "what do"])
1733    {
1734        Some(DirectAnswerKind::GemmaNative)
1735    } else if lower.contains("verify_build")
1736        && lower.contains(".hematite/settings.json")
1737        && contains_any(
1738            &lower,
1739            &["build", "test", "lint", "fix", "verification commands"],
1740        )
1741    {
1742        Some(DirectAnswerKind::VerifyProfiles)
1743    } else if (lower.contains("carry forward by default")
1744        || lower.contains("session memory should you carry forward")
1745        || (lower.contains("carry forward")
1746            && contains_any(
1747                &lower,
1748                &[
1749                    "besides the active task",
1750                    "blocker",
1751                    "compacts",
1752                    "recovers from a blocker",
1753                    "session state",
1754                ],
1755            )))
1756        && contains_any(
1757            &lower,
1758            &[
1759                "restarted hematite",
1760                "restarted",
1761                "avoid carrying forward",
1762                "session state",
1763                "active task",
1764                "blocker",
1765                "compacts",
1766                "recovers from a blocker",
1767            ],
1768        )
1769    {
1770        Some(DirectAnswerKind::SessionMemory)
1771    } else if contains_any(
1772        &lower,
1773        &[
1774            "recovery recipe",
1775            "recovery recipes",
1776            "recovery step",
1777            "recovery steps",
1778        ],
1779    ) && contains_any(
1780        &lower,
1781        &[
1782            "blocker",
1783            "runtime failure",
1784            "degrades",
1785            "context window",
1786            "context-window",
1787            "operator",
1788        ],
1789    ) {
1790        Some(DirectAnswerKind::RecoveryRecipes)
1791    } else if !architecture_overview_mode
1792        && contains_any(
1793            &lower,
1794            &[
1795                "mcp server health",
1796                "mcp runtime state",
1797                "mcp lifecycle",
1798                "mcp state",
1799                "mcp healthy",
1800                "mcp degraded",
1801                "mcp failed",
1802            ],
1803        )
1804    {
1805        Some(DirectAnswerKind::McpLifecycle)
1806    } else if contains_any(
1807        &lower,
1808        &[
1809            "allowed, denied, or require approval",
1810            "allowed denied or require approval",
1811            "allow, ask, or deny",
1812            "tool call should be allowed",
1813            "authorization logic",
1814            "workspace trust",
1815            "trust-allowlisted",
1816        ],
1817    ) {
1818        Some(DirectAnswerKind::AuthorizationPolicy)
1819    } else if contains_any(
1820        &lower,
1821        &[
1822            "tool classes",
1823            "tool class",
1824            "flat tool list",
1825            "runtime tool classes",
1826            "different runtime tool classes",
1827        ],
1828    ) || (lower.contains("repo reads")
1829        && lower.contains("repo writes")
1830        && contains_any(
1831            &lower,
1832            &[
1833                "verification tools",
1834                "git tools",
1835                "external mcp tools",
1836                "different runtime",
1837            ],
1838        ))
1839    {
1840        Some(DirectAnswerKind::ToolClasses)
1841    } else if contains_any(
1842        &lower,
1843        &[
1844            "built-in tool catalog",
1845            "builtin tool catalog",
1846            "builtin-tool dispatch",
1847            "built-in tool dispatch",
1848            "tool registry ownership",
1849            "which file now owns",
1850        ],
1851    ) && contains_any(
1852        &lower,
1853        &[
1854            "tool catalog",
1855            "dispatch path",
1856            "dispatch",
1857            "tool registry",
1858            "owns",
1859        ],
1860    ) {
1861        Some(DirectAnswerKind::ToolRegistryOwnership)
1862    } else if (lower.contains("other coding languages")
1863        || lower.contains("what languages")
1864        || lower.contains("know other languages"))
1865        && contains_any(
1866            &lower,
1867            &[
1868                "capable of making projects",
1869                "can you make projects",
1870                "can you build projects",
1871            ],
1872        )
1873    {
1874        Some(DirectAnswerKind::LanguageCapability)
1875    } else if workflow_mode == WorkflowMode::Architect
1876        && (lower.contains("session reset")
1877            || (lower.contains("/clear") && lower.contains("/new") && lower.contains("/forget")))
1878        && contains_any(&lower, &["redesign", "clearer", "easier", "understand"])
1879    {
1880        Some(DirectAnswerKind::ArchitectSessionResetPlan)
1881    } else if toolchain_mode
1882        && lower.contains("read-only")
1883        && contains_any(
1884            &lower,
1885            &[
1886                "tooling discipline",
1887                "investigation plan",
1888                "best read-only toolchain",
1889                "tool choice discipline",
1890                "what tools would you choose first",
1891            ],
1892        )
1893    {
1894        Some(DirectAnswerKind::Toolchain)
1895    } else {
1896        None
1897    };
1898
1899    let primary_class = if direct_answer.is_some()
1900        || mentions_stable_product_surface(&lower)
1901        || mentions_product_truth_routing(&lower)
1902    {
1903        QueryIntentClass::ProductTruth
1904    } else if architecture_overview_mode {
1905        QueryIntentClass::RepoArchitecture
1906    } else if toolchain_mode {
1907        QueryIntentClass::Toolchain
1908    } else if capability_mode {
1909        QueryIntentClass::Capability
1910    } else if mentions_runtime_trace || anti_guess || lower.contains("read-only") {
1911        QueryIntentClass::RuntimeDiagnosis
1912    } else if looks_like_mutation_request(user_input) {
1913        QueryIntentClass::Implementation
1914    } else {
1915        QueryIntentClass::Unknown
1916    };
1917
1918    QueryIntent {
1919        primary_class,
1920        direct_answer,
1921        grounded_trace_mode: mentions_runtime_trace || lower.contains("read-only") || anti_guess,
1922        capability_mode,
1923        capability_needs_repo,
1924        toolchain_mode,
1925        host_inspection_mode,
1926        maintainer_workflow_mode,
1927        workspace_workflow_mode,
1928        architecture_overview_mode,
1929    }
1930}
1931
1932pub(crate) fn is_capability_probe_tool(name: &str) -> bool {
1933    matches!(
1934        name,
1935        "read_file"
1936            | "inspect_lines"
1937            | "list_files"
1938            | "grep_files"
1939            | "lsp_definitions"
1940            | "lsp_references"
1941            | "lsp_hover"
1942            | "lsp_search_symbol"
1943            | "lsp_get_diagnostics"
1944            | "trace_runtime_flow"
1945            | "auto_pin_context"
1946            | "list_pinned"
1947    )
1948}
1949
1950/// Returns true when the user's query involves computation that must be exact —
1951/// checksums, financial math, statistics, date arithmetic, algorithmic verification, etc.
1952/// Used by the harness to inject a pre-turn nudge toward run_code instead of model memory.
1953pub fn needs_computation_sandbox(user_input: &str) -> bool {
1954    let lower = user_input.to_lowercase();
1955    let hash_or_checksum = lower.contains("sha")
1956        || lower.contains("md5")
1957        || lower.contains("checksum")
1958        || lower.contains("crc")
1959        || lower.contains("hash")
1960        || lower.contains("fingerprint");
1961    let financial =
1962        (lower.contains("calculat") || lower.contains("compute") || lower.contains("what is"))
1963            && (lower.contains("percent")
1964                || lower.contains("%")
1965                || lower.contains("interest")
1966                || lower.contains("compound")
1967                || lower.contains("roi")
1968                || lower.contains("tax")
1969                || lower.contains("discount")
1970                || lower.contains("profit")
1971                || lower.contains("loss"));
1972    let statistics = lower.contains("standard deviation")
1973        || lower.contains("std dev")
1974        || lower.contains("mean of")
1975        || lower.contains("median of")
1976        || lower.contains("average of")
1977        || lower.contains("variance")
1978        || lower.contains("regression")
1979        || lower.contains("correlation");
1980    let date_math = (lower.contains("how many days")
1981        || lower.contains("days between")
1982        || lower.contains("days until")
1983        || lower.contains("days since")
1984        || lower.contains("unix timestamp")
1985        || lower.contains("epoch")
1986        || lower.contains("time zone")
1987        || lower.contains("timezone"))
1988        && (lower.contains("date")
1989            || lower.contains("day")
1990            || lower.contains("timestamp")
1991            || lower.contains("time"));
1992    let algorithmic = lower.contains("is prime")
1993        || lower.contains("prime number")
1994        || lower.contains("factori")
1995        || lower.contains("fibonacci")
1996        || lower.contains("factorial")
1997        || lower.contains("sort this")
1998        || lower.contains("verify this algorithm")
1999        || lower.contains("run this code")
2000        || lower.contains("execute this");
2001    let unit_conversion = (lower.contains("convert") || lower.contains("how many"))
2002        && (lower.contains(" bytes")
2003            || lower.contains(" kb")
2004            || lower.contains(" mb")
2005            || lower.contains(" gb")
2006            || lower.contains(" tb")
2007            || lower.contains("gigabyte")
2008            || lower.contains("megabyte")
2009            || lower.contains("celsius")
2010            || lower.contains("fahrenheit")
2011            || lower.contains("kelvin")
2012            || lower.contains("kilometers")
2013            || lower.contains("miles")
2014            || lower.contains("pounds")
2015            || lower.contains("kilograms"));
2016    hash_or_checksum || financial || statistics || date_math || algorithmic || unit_conversion
2017}
2018
2019#[cfg(test)]
2020mod tests {
2021    use super::*;
2022
2023    #[test]
2024    fn classify_query_intent_routes_creator_questions_to_about() {
2025        let intent = classify_query_intent(WorkflowMode::Auto, "Who created Hematite?");
2026        assert_eq!(intent.direct_answer, Some(DirectAnswerKind::About));
2027
2028        let intent = classify_query_intent(WorkflowMode::Auto, "/about");
2029        assert_eq!(intent.direct_answer, Some(DirectAnswerKind::About));
2030    }
2031
2032    #[test]
2033    fn classify_query_intent_marks_maintainer_workflow_requests() {
2034        let intent = classify_query_intent(
2035            WorkflowMode::Auto,
2036            "Run my cleanup scripts and prune old artifacts.",
2037        );
2038        assert!(intent.maintainer_workflow_mode);
2039        assert_eq!(
2040            preferred_maintainer_workflow("Rebuild the local portable and update PATH."),
2041            Some("package_windows")
2042        );
2043        assert_eq!(
2044            preferred_maintainer_workflow("Run the release flow and publish crates."),
2045            Some("release")
2046        );
2047    }
2048
2049    #[test]
2050    fn classify_query_intent_marks_workspace_workflow_requests() {
2051        let intent = classify_query_intent(WorkflowMode::Auto, "Run the tests in this project.");
2052        assert!(intent.workspace_workflow_mode);
2053        assert_eq!(
2054            preferred_workspace_workflow("Run the tests in this project."),
2055            Some("test")
2056        );
2057        assert_eq!(
2058            preferred_workspace_workflow("Run npm run dev in this repo."),
2059            Some("script")
2060        );
2061    }
2062}