Coordinate parallel reviewers to assess ADRs and enforce blocking gates.
npx skills add https://github.com/notque/vexjoy-agent --skill adr-consultation
Please help me install this Agent Skill. Skill: adr-consultation Source: https://github.com/notque/vexjoy-agent/tree/main/skills/process/adr-consultation Command: npx skills add https://github.com/notque/vexjoy-agent --skill adr-consultation
This Skill prevents architecture decisions from moving forward without challenging assumptions, assessing user impact, and checking system-level risks before implementation.
severity: blocking issue results in a BLOCKED verdict, preventing rationalization-driven “skip” behavior.Run the adr-consultation Skill with your intent like “consult on ADR <name or path> before implementing this architecture change.”
High-intent search queries and answers about installing and using this skill.
To review an Architecture Decision Record before implementation, dispatch parallel multi-agent consultations that evaluate premises, user impact, and system health, synthesizing concerns into a unified PROCEED or BLOCKED verdict.
Multi-agent consultation for architecture decisions dispatches specialized reviewers to evaluate an ADR from contrarian, user-advocate, and meta-process perspectives, producing lens-specific verdicts synthesized into a final decision.
Yes, you can enforce a hard gate on blocking concerns during architecture review. Any severity blocking issue automatically results in a BLOCKED verdict, preventing rationalization-driven skip behavior.
Synthesize architecture review feedback into a reproducible format by writing each reviewer's output to disk as artifacts, then consolidating concerns into concerns.md and synthesis.md files for a unified verdict.
You need to consult on an Architecture Decision Record before implementing Medium or larger architecture changes, such as adding a new subsystem, to challenge assumptions and reduce costly post-implementation rework.
Architecture decision gating works without external dependencies, using dispatched Task calls to write reviewer artifacts to disk and synthesize concerns internally for a final PROCEED or BLOCKED verdict.
pluginagentmarketplaceGenerate technical decision documents with evaluation frameworks and action plans.
machenjieConvert ambiguous product-change requests into decision-ready clarification records.
MicrockCoordinate multi-agent swarms for parallel task execution across mesh, hierarchical, and adaptive topologies.
xxihDiscover, select, and run multiple AI agents in parallel with aggregated results.