メニュー
Collect bounded HTTP metadata, response evidence, and verified technology fingerprints from Web services in an explicitly authorized CyberEdge scope.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
SOC 職業分類に基づく
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/Symph0nia/CyberEdge --skill cyberedge-observe-websitesコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
SKILL.md を表示中
Run or continuously monitor one evidence-backed CyberEdge assessment workflow for an explicitly authorized Scope. Use when an AI operator must check readiness, expand domain assets, inventory services and websites, run bounded vulnerability and intelligence stages, follow one Task, create an explicitly requested recurring assessment, inspect exposure changes, and report exact coverage without claiming unavailable adapters were completed.
Correlate exact, evidence-backed CPE identities already observed by CyberEdge with normalized NVD CVE records. Use when an AI operator must run CVE intelligence for an authorized Scope, interpret CVSS metadata and lifecycle state, or distinguish completed CPE coverage from missing identity or provider failure without guessing products or versions.
Search bounded GitHub public-code metadata for exact domain names in an explicitly authorized CyberEdge Scope. Use when an AI operator must run public-code intelligence, inspect evidence-backed reference candidates, or distinguish completed negative coverage from provider failure without retrieving source content or claiming confirmed credential leakage.
| name | cyberedge-observe-websites |
| description | Collect bounded HTTP metadata, response evidence, and verified technology fingerprints from Web services in an explicitly authorized CyberEdge scope. |
Use cyberedge-agent; this machine bridge is not a human CLI.
scan.active and website.read.policy_service_baseline for the authorized Scope.search_websites for the same Scope.fingerprints[].evidence_id with http.response Observations and TaskReport Evidence.discovered_paths and correlate http.crawl Observations with their retained Evidence.screenshot_evidence_id is present, treat it as an offline rendering of retained HTML, not proof of live browser behavior.http-host-collision-v1 Findings with http.host_collision_check Observations. Require the comparison Evidence to contain both the direct-address baseline and Host-routed response; do not infer a collision from a different body hash alone.server only as a header-derived hint. Treat a structured fingerprint as verified only for its versioned detector rule; do not infer additional products or versions.For recurring monitoring, create a Schedule with the same active policy and call search_exposure_changes using its schedule_id. Treat APPEARED, DISAPPEARED, and MODIFIED as deterministic diffs between successful Task snapshots. Collection errors suppress disappearance events.
The server probes only open 80, 443, 8080, and 8443, never follows redirects, enforces a five-second request timeout and rejects response bodies over 1 MiB. http.error means collection failed; it does not prove that no website exists.
Current strong-signature rules identify WordPress and Grafana from retained HTML. Generic product-name text is insufficient. WordPress versions are emitted only when an explicit generator value supplies one.
The crawler is depth one and accepts at most 16 same-origin absolute paths found on the root page. It rejects query strings, scheme-relative/cross-origin links, traversal segments, encoded traversal, and caller-supplied paths. Do not describe this as a general-purpose spider.
Host collision checks run only within an active Scope containing domains and alternate IP assets. A domain's current DNS addresses are excluded, candidates are bounded server-side, and probe failures appear as http.host_collision_error; treat those failures as unknown coverage, not a negative result.
{"request_id":"req-1","idempotency_key":"idem-1","agent_id":"codex-main","skill_name":"cyberedge-observe-websites","skill_version":"0.1.0","action":"search_websites","scope_id":"scope-id"}