subagent-hunt-orchestration

Coordinates specialist subagents to split offensive security campaigns across parallel investigation surfaces.

7|4|Updated Jun 22, 2026
One-click install
npx skills add https://github.com/dbx0/skills --skill subagent-hunt-orchestration-dbx0
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-hunt-orchestration
Source: https://github.com/dbx0/skills/tree/main/skills/methodology/orchestration/subagent-hunt-orchestration
Command: npx skills add https://github.com/dbx0/skills --skill subagent-hunt-orchestration-dbx0

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Large offensive security engagements overwhelm a single linear agent loop, causing duplicated work, missed surfaces, and stalled hypothesis trees when multiple branches are live at once. ## Core Features & Use Cases - Specialist Roster: Defines default subagent roles including surface mapper, business-logic hunter, GraphQL/API hunter, import-parser hunter, chain builder, and report synthesizer. - Structured Brief Format: Every subagent task includes target scope, dossier summary, one surface family, one success condition, and one stop condition to keep work bounded. - Deconfliction and Merge Discipline: Prevents two subagents from chasing the same surface and requires dossier updates, disproven paths, and chain candidates after each return. - Use Case: During an assessment of a large SaaS target, dispatch a surface mapper to enumerate object families while a parser hunter stresses upload/import paths, then merge results to decide whether to deepen, pivot, or report. ## Quick Start Ask the agent to split the current engagement across specialist subagents using the brief format with defined success and stop conditions for each surface.

Frequently Asked Questions about subagent-hunt-orchestration

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I split a penetration test across multiple AI subagents?

Assign each subagent one surface family with a structured brief containing target scope, dossier summary, a success condition, and a stop condition. Default roles include surface mapper, business-logic hunter, API hunter, parser hunter, chain builder, and report synthesizer.

When should I use subagent orchestration instead of a single agent loop?

Use it when the target is broad, the workflow is deep, or the hypothesis tree has multiple live branches. A single linear loop wastes time on large targets where parallel surface investigation is possible.

How do I prevent subagents from duplicating work?

Apply deconfliction by assigning distinct roles per surface: the mapper finds, the manual hunter reasons, the API hunter compares transports, the parser hunter stresses ingestion, and the chain builder composes. Never send two subagents after the same surface with the same goal.

What should a subagent brief contain for security testing?

Each brief needs the target scope, a current dossier summary, one surface family, one success condition, and one stop condition. For example, a parser hunter stops on one validated primitive or three disproven branches.

Why does subagent output sometimes fail to improve an engagement?

If returned output does not change campaign state, the brief was too vague. Merge discipline requires updating the dossier, recording disproven paths, adding chain candidates, and deciding whether to deepen, pivot, or report after each return.