meridianai-phaselisted

# /meridian:ai-phase — AI Integration Phase Type Spawns 3 parallel subagents (domain researcher, framework selector, eval planner) for a phase that integrates an LLM, AI model, or ML component. Produces `AI-SPEC.md` in the phase artifact directory. The plan phase soft-gates on this artifact for AI phases. **Position in workflow:** `ai-phase → spec-phase → discuss-phase → plan-phase → execute-phase` ## Arguments - (no args) — spec the current pending/planned phase - `--phase <id>` — specify a phase by ID - `--skip-eval` — skip the eval planner subagent (faster, 2 subagents only) - `--skip-ai` — bypass gate warning in /meridian:plan (emergency only) ## Keywords ai, llm, model, claude, openai, anthropic, gpt, embeddings, rag, agents, tool use, prompt engineering, eval, evaluation, guardrails, inference, fine-tuning, pre-plan ## Procedure ### Step 1: Find Target Phase ```bash PYTHONPATH=$MERIDIAN_HOME uv run --project $MERIDIAN_HOME -- python -c " import json from scripts.db import connect, get_db_path from scripts.ai_phase import get_ai_context conn = connect(get_db_path('.')) ctx = get_ai_context(conn, phase_id=<phase_id_or_None>) print(json.dumps(ctx, indent=2, default=str)) conn.close() " ``` Pass the `--phase <id>` value as `phase_id`, or `None` if not specified. If result contains `"error"`, display it and stop — tell the user to run `/meridian:plan` first. Store: `phase_id`, `phase_name`, `description`, `acceptance_criteria`, `tech_stack`, `phase_dir`, `slug`.