SKILL.md

GitHub
namebloodhound-query
descriptionUse when the user explicitly wants BloodHound Cypher or query work: write a BloodHound query, review or explain Cypher, optimize a query, adapt a saved query, validate query safety or performance, convert a graph question into query syntax, or design follow-up graph questions across BloodHound CE/BHE, AzureHound, OpenHound GitHub, OpenHound Jamf, OpenHound Okta, or other OpenGraph data. Do not use as the default entry skill for generic path triage, connection checks, or AD/Azure/OpenHound exposure analysis when no explicit query-authoring task is requested.
licenseMIT
metadata{"author":"turbo","version":"0.1.0","category":"security"}

BloodHound Query Workflow

Use this as the shared query authoring/review workflow across BloodHound, AzureHound, OpenHound GitHub, OpenHound Jamf, OpenHound Okta, and custom OpenGraph graphs.

Direct triggers

Use this skill when the task mentions any of the following:

  • write a BloodHound query
  • write Cypher
  • explain this Cypher
  • review this query
  • optimize this BloodHound query
  • adapt this saved query
  • turn this path question into Cypher
  • validate this BloodHound query

Route to instead

  • Use $bloodhound-analysis for generic BloodHound asks or initial graph triage.
  • Use the domain skill when the path question is clear and the user is asking for findings rather than query authoring: $bloodhound-ad-analysis, $azurehound-analysis, $openhound-github, $openhound-jamf, or $openhound-okta.

Required context

  • Confirm the assessment/lab is authorized and in scope.
  • Identify the graph domain and available collectors/extensions.
  • If live BloodHound MCP access is unavailable, produce a query/workflow plan and clearly avoid claiming observed graph facts.

Workflow

  1. Read ../../references/docs/bloodhound-query-methodology.md.
  2. Choose the domain skill when the graph is known: $bloodhound-ad-analysis, $azurehound-analysis, $openhound-github, $openhound-jamf, or $openhound-okta.
  3. Use ../../references/docs/source-index.md to locate the matching query index and snapshots.
  4. Adapt a saved-query pattern first; only invent a new query when no pattern fits.
  5. For OpenGraph work, inspect ../../references/examples/example-cypher.md and ../../references/examples/node-edge-reference.md before inventing labels or edge kinds.
  6. For SCIM/hybrid identity work, read ../../references/docs/scim-methodology.md and document each bridge edge explicitly.
  7. Keep the query read-only, bounded, label-specific, and explicit about relationship direction.
  8. Return the query with parameters, expected result shape, analysis guidance, caveats, and next queries.

Output contract

  • Query
  • Parameters to replace
  • Purpose
  • Expected result shape
  • Analysis guidance
  • Caveats / data-quality assumptions
  • Next queries