一键导入
monitoring
Turn investigation findings into explicit, user-approved monitoring recommendations. Spotlight recommends; Mycroft owns any Scoutpost action.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Turn investigation findings into explicit, user-approved monitoring recommendations. Spotlight recommends; Mycroft owns any Scoutpost action.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when an investigation step may need an external integration such as browser acquisition, Maigret account discovery, Junkipedia narrative tracking, OSINT Navigator tool discovery, Noosphere C2PA signing, or Unpaywall access lookup.
OSINT investigation orchestrator — guides verified investigations from lead to findings to knowledge ingestion. Triggers on "investigate", "investigation", "OSINT", "look into", "dig into".
OSINT investigation toolkit — local SQL index of 12,500 tools (`osint-tools find`, offline), methodology guides, and optional OSINT Navigator (subscription tier). Works offline with any LLM.
Portable claim-to-evidence grounding for knowledge work, fact-checking, and Spotlight investigations. Use when extracting claims or findings, assigning confidence, diagnosing weak support, or deciding whether material is a lead, partially grounded claim, verified finding, disputed claim, or false claim.
Portable safe command construction for any agent skill or recipe. Use before a shell call containing user, model, scraped, configuration, filesystem, or generated values.
Stage and commit approved investigation findings through the Knowledge Workspace Port into the spotlight namespace.
| name | monitoring |
| description | Turn investigation findings into explicit, user-approved monitoring recommendations. Spotlight recommends; Mycroft owns any Scoutpost action. |
| version | 3.0 |
| invocable_by | ["orchestrator"] |
| requires | [] |
Spotlight owns active casework, not persistent monitoring. It may identify
targets worth watching and record them in monitoring_recommendations[] in
data/findings.json. It never installs Scoutpost, reads its credentials,
creates a scout, queries a scout, or stores Scoutpost identifiers in a case.
Use one clear handoff, never a direct Scoutpost command:
"This investigation suggests monitoring {target} for {criteria}. Would you like to hand this to Mycroft for monitoring setup?"
When the answer is yes, include the target, rationale, cadence preference, and case reference in the Mycroft prompt. The Mycroft skill owns authentication, Scoutpost CLI/API selection, scout creation, and durable monitoring state.
monitoring_recommendations[] is evidence-adjacent case context, not a live
monitor registry. Keep only the recommendation and the user's decision:
{
"target": "Example organisation",
"priority": "high",
"rationale": "New procurement notices could change the finding.",
"criteria": "new contracts, ownership changes, enforcement action",
"status": "recommended|handed_to_mycroft|declined"
}
Do not write API keys, project IDs, scout IDs, remote-monitor results, or runtime task handles into Spotlight case state.
In sensitive mode, record recommendations only. Do not hand anything to an external monitoring system unless the user explicitly resumes that choice.