Orchestrate structured multi-agent debates to generate Architecture Decision Records.
npx skills add https://github.com/dnhess/spectra --skill decision-board
Please help me install this Agent Skill. Skill: decision-board Source: https://github.com/dnhess/spectra/tree/main/decision-board Command: npx skills add https://github.com/dnhess/spectra --skill decision-board
Orchestrates a team of expert agents who conduct a structured, multi-perspective debate on an architectural or technical decision using a blackboard architecture. Agents write structured JSON files to a shared session directory; the moderator polls for file existence, reads results, and writes all events to the JSONL log. Debates unfold with stances, challenges, concessions, and a final Architecture Decision Record (ADR).
Initiate a session by providing a decision question and optional options, then let the moderator orchestrate the debate.
High-intent search queries and answers about installing and using this skill.
To automate an architectural decision debate and generate an ADR, initiate a session with your decision question and optional constraints. A moderator agent orchestrates a multi-perspective debate, logging JSONL events and outputting a final Architecture Decision Record.
A blackboard architecture for technical decisions uses a shared session directory where expert agents write structured JSON files. A moderator polls for these files, reads results, and writes events to a JSONL log to orchestrate a reproducible multi-perspective debate.
You can evaluate migration plans and technology choices by running a structured debate with role-based personas like Architect, Pragmatist, and Economist. Agents exchange stances, challenges, and concessions to deliver a well-supported recommendation.
Yes, the structured debate supports quick evaluations for fast technical decisions through tiered modes. You can select Quick, Standard, or Deep debate modes to match your context budget constraints and desired depth of analysis.
No specific dependencies are required to run a session-based architectural decision debate. The process relies on a fixed blackboard-like session protocol, writing structured JSON files to a shared directory and logging all events to JSONL.
You should not use an automated multi-agent debate when context budget is severely limited or the decision requires immediate resolution. The protocol involves multiple agents writing JSON files and logging JSONL events, which demands processing overhead.
Ramprasad4121Ask targeted questions to clarify objectives, scope, constraints, and environment.
0xHarbsGenerate session progress logs and update project documentation files.
rdmptvManage enterprise session state and token budgets for MoAI-ADK workflows.
tmalcolm-0607Run structured multi-agent adversarial debates with configurable formats and verdict generation.
