feat(hunter): add standard-mode scheduling

src/openultrasast/cli.py

@@ -7,6 +7,7 @@ from pathlib import Path
 from .config import load_config
 from .findings import quick_scan_findings, write_findings
 from .harness import HarnessRuntime, HarnessTraceWriter, write_harness_config
+from .hunter import run_hunter_pool, write_hunter_trajectories
 from .index import build_code_chunks
 from .mapping import analyze_entry_points, attach_reachability_hints, ingest_sarif, write_entry_points, write_static_hints
 from .preprocess import preprocess_repository, write_preprocess_artifact
@@ -40,8 +41,8 @@ def main(argv: list[str] | None = None) -> int:
 
 
 def _scan(path: Path, config_path: Path, mode: str, fail_on: str) -> int:
-    if mode != "quick":
-        raise SystemExit(f"mode {mode!r} is specified but only quick preprocess artifacts are implemented")
+    if mode == "deep":
+        raise SystemExit("mode 'deep' is specified but sandboxed dynamic analysis is not implemented")
     if not path.exists() or not path.is_dir():
         raise SystemExit(f"scan path is not a directory: {path}")
 
@@ -66,13 +67,22 @@ def _scan(path: Path, config_path: Path, mode: str, fail_on: str) -> int:
     write_preprocess_artifact(snapshot, targets, run.root / "preprocess" / "file_targets.json")
     rankings = runtime.run_stage("rank", lambda: rank_targets(targets))
     write_rankings(rankings, run.root / "rank" / "ranking.json")
-    findings = runtime.run_stage("quick_findings", lambda: quick_scan_findings(run.target, targets, rankings))
+    if mode == "standard":
+        hunter_result = runtime.run_stage("hunter_pool", lambda: run_hunter_pool(run.target, targets, rankings, scan_id=run.scan_id))
+        findings = hunter_result.findings
+        trajectories = hunter_result.trajectories
+    else:
+        findings = runtime.run_stage("quick_findings", lambda: quick_scan_findings(run.target, targets, rankings))
+        trajectories = []
     findings_path = run.root / "findings.json"
     verification_path = run.root / "verification.json"
     markdown_path = run.root / "report.md"
     sarif_path = run.root / "report.sarif"
     manifest_path = run.root / "manifest.json"
+    trajectories_path = run.root / "traces" / "hunter_trajectories.jsonl"
     write_findings(findings, findings_path)
+    if trajectories:
+        write_hunter_trajectories(trajectories, trajectories_path)
     verifications = runtime.run_stage("verify", lambda: verify_findings(findings))
     write_verification_results(verifications, verification_path)
     runtime.run_stage("report", lambda: write_markdown_report(findings, markdown_path, verifications))
@@ -83,12 +93,13 @@ def _scan(path: Path, config_path: Path, mode: str, fail_on: str) -> int:
             run=run,
             findings=findings,
             verifications=verifications,
-            artifact_paths={
-                "findings": findings_path,
-                "verification": verification_path,
-                "markdown": markdown_path,
-                "sarif": sarif_path,
-            },
+            artifact_paths=_artifact_paths(
+                findings=findings_path,
+                verification=verification_path,
+                markdown=markdown_path,
+                sarif=sarif_path,
+                trajectories=trajectories_path if trajectories else None,
+            ),
             path=manifest_path,
         ),
     )
@@ -110,6 +121,10 @@ def _load_static_hints(sarif_paths: tuple[str, ...]) -> list[object]:
     return hints
 
 
+def _artifact_paths(**paths: Path | None) -> dict[str, Path]:
+    return {name: path for name, path in paths.items() if path is not None}
+
+
 def _index(path: Path, config_path: Path, chunk_lines: int) -> int:
     if not path.exists() or not path.is_dir():
         raise SystemExit(f"index path is not a directory: {path}")

src/openultrasast/hunter.py

@@ -0,0 +1,161 @@
+from __future__ import annotations
+
+import json
+from concurrent.futures import ThreadPoolExecutor
+from dataclasses import asdict, dataclass
+from pathlib import Path
+from typing import Literal
+
+from .findings import StaticFinding, quick_scan_findings
+from .preprocess import FileTarget
+from .rank import RankingScore
+
+HunterTier = Literal["A", "B", "C"]
+
+
+@dataclass(frozen=True)
+class HunterBudget:
+    tier: HunterTier
+    max_targets: int
+    max_findings_per_target: int
+    retrieval_budget_chars: int
+    skill_budget_chars: int
+
+
+@dataclass(frozen=True)
+class HunterTask:
+    target: FileTarget
+    ranking: RankingScore
+    tier: HunterTier
+    budget: HunterBudget
+    selected_skills: list[str]
+    retrieval_context: str
+
+
+@dataclass(frozen=True)
+class HunterTrajectory:
+    scan_id: str
+    target_path: str
+    tier: HunterTier
+    priority: float
+    selected_skills: list[str]
+    retrieval_context_chars: int
+    findings: list[str]
+    status: str
+
+
+@dataclass(frozen=True)
+class HunterPoolResult:
+    findings: list[StaticFinding]
+    trajectories: list[HunterTrajectory]
+
+
+DEFAULT_BUDGETS = {
+    "A": HunterBudget(tier="A", max_targets=20, max_findings_per_target=20, retrieval_budget_chars=6000, skill_budget_chars=2500),
+    "B": HunterBudget(tier="B", max_targets=40, max_findings_per_target=10, retrieval_budget_chars=4000, skill_budget_chars=1500),
+    "C": HunterBudget(tier="C", max_targets=80, max_findings_per_target=5, retrieval_budget_chars=2000, skill_budget_chars=800),
+}
+
+
+def assign_tier(priority: float) -> HunterTier:
+    if priority >= 3:
+        return "A"
+    if priority >= 2:
+        return "B"
+    return "C"
+
+
+def schedule_hunter_tasks(
+    targets: list[FileTarget],
+    rankings: list[RankingScore],
+    *,
+    retrieval_context_by_path: dict[str, str] | None = None,
+) -> list[HunterTask]:
+    target_by_path = {target.path: target for target in targets}
+    retrieval_context_by_path = retrieval_context_by_path or {}
+    scheduled: list[HunterTask] = []
+    per_tier_counts: dict[HunterTier, int] = {"A": 0, "B": 0, "C": 0}
+
+    for ranking in sorted(rankings, key=lambda item: (-item.priority, item.path)):
+        target = target_by_path.get(ranking.path)
+        if target is None:
+            continue
+        tier = assign_tier(ranking.priority)
+        budget = DEFAULT_BUDGETS[tier]
+        if per_tier_counts[tier] >= budget.max_targets:
+            continue
+        per_tier_counts[tier] += 1
+        scheduled.append(
+            HunterTask(
+                target=target,
+                ranking=ranking,
+                tier=tier,
+                budget=budget,
+                selected_skills=select_skill_snippets(target),
+                retrieval_context=retrieval_context_by_path.get(target.path, "")[: budget.retrieval_budget_chars],
+            )
+        )
+    return scheduled
+
+
+def run_hunter_pool(root: Path, targets: list[FileTarget], rankings: list[RankingScore], *, scan_id: str) -> HunterPoolResult:
+    tasks = schedule_hunter_tasks(targets, rankings, retrieval_context_by_path=build_retrieval_context_by_path(root, targets))
+    findings: list[StaticFinding] = []
+    trajectories: list[HunterTrajectory] = []
+    max_workers = min(8, max(1, len(tasks)))
+    with ThreadPoolExecutor(max_workers=max_workers) as executor:
+        for task_findings, trajectory in executor.map(lambda task: _run_hunter_task(root, scan_id, task), tasks):
+            findings.extend(task_findings)
+            trajectories.append(trajectory)
+    return HunterPoolResult(findings=sorted(findings, key=lambda item: item.finding_id), trajectories=trajectories)
+
+
+def write_hunter_trajectories(trajectories: list[HunterTrajectory], path: Path) -> None:
+    path.parent.mkdir(parents=True, exist_ok=True)
+    path.write_text("".join(json.dumps(asdict(trajectory), sort_keys=True) + "\n" for trajectory in trajectories))
+
+
+def build_retrieval_context_by_path(root: Path, targets: list[FileTarget], *, max_chars: int = 6000) -> dict[str, str]:
+    contexts: dict[str, str] = {}
+    for target in targets:
+        source = _read_source(root / target.path, max_chars=max_chars)
+        if source:
+            contexts[target.path] = f"[repo_code] {target.path}\n{source}"
+    return contexts
+
+
+def select_skill_snippets(target: FileTarget, *, max_snippets: int = 4) -> list[str]:
+    tags = set(target.tags)
+    snippets: list[str] = []
+    if target.language in {"c", "cpp", "c++"} or "memory_unsafe" in tags:
+        snippets.extend(["address-sanitizer", "libfuzzer", "harness-writing"])
+    if "parser" in tags or "fuzzable" in tags:
+        snippets.extend(["fuzzing-dictionary", "fuzzing-obstacles"])
+    if "crypto" in tags:
+        snippets.extend(["constant-time-analysis", "wycheproof"])
+    if "auth_boundary" in tags or "network_entry" in tags:
+        snippets.extend(["sharp-edges", "insecure-defaults"])
+    if "deserialization" in tags:
+        snippets.append("semgrep-rule-creator")
+    return sorted(dict.fromkeys(snippets))[:max_snippets]
+
+
+def _run_hunter_task(root: Path, scan_id: str, task: HunterTask) -> tuple[list[StaticFinding], HunterTrajectory]:
+    findings = quick_scan_findings(root, [task.target], [task.ranking])[: task.budget.max_findings_per_target]
+    return findings, HunterTrajectory(
+        scan_id=scan_id,
+        target_path=task.target.path,
+        tier=task.tier,
+        priority=task.ranking.priority,
+        selected_skills=task.selected_skills,
+        retrieval_context_chars=len(task.retrieval_context),
+        findings=[finding.finding_id for finding in findings],
+        status="completed",
+    )
+
+
+def _read_source(path: Path, *, max_chars: int) -> str:
+    try:
+        return path.read_text(errors="ignore")[:max_chars]
+    except OSError:
+        return ""

src/openultrasast/reports.py

@@ -88,12 +88,7 @@ def write_manifest(
                 "verification_status": verification_by_id[finding.finding_id].status
                 if finding.finding_id in verification_by_id
                 else "not_run",
-                "artifact_refs": {
-                    "findings_json": _relative_artifact(run.root, artifact_paths["findings"]),
-                    "verification_json": _relative_artifact(run.root, artifact_paths["verification"]),
-                    "markdown": _relative_artifact(run.root, artifact_paths["markdown"]),
-                    "sarif": _relative_artifact(run.root, artifact_paths["sarif"]),
-                },
+                "artifact_refs": _artifact_refs(run.root, artifact_paths),
             }
             for finding in findings
         ],
@@ -170,6 +165,18 @@ def _verification_by_id(verifications: list[VerificationResult]) -> dict[str, Ve
     return {result.finding_id: result for result in verifications}
 
 
+def _artifact_refs(root: Path, artifact_paths: dict[str, Path]) -> dict[str, str]:
+    refs = {
+        "findings_json": _relative_artifact(root, artifact_paths["findings"]),
+        "verification_json": _relative_artifact(root, artifact_paths["verification"]),
+        "markdown": _relative_artifact(root, artifact_paths["markdown"]),
+        "sarif": _relative_artifact(root, artifact_paths["sarif"]),
+    }
+    if "trajectories" in artifact_paths:
+        refs["trajectories_jsonl"] = _relative_artifact(root, artifact_paths["trajectories"])
+    return refs
+
+
 def _relative_artifact(root: Path, path: Path) -> str:
     try:
         return path.relative_to(root).as_posix()