AVIDS2

Use when Claude Code needs Memorix shared memory, reasoning, Git Memory, mini-skills, session handoff, orchestration coordination, or integration troubleshooting.

Use when prior workspace context, past decisions, solved bugs, handoff state, or durable project knowledge would help a coding task.

Use when prior workspace context, past decisions, solved bugs, handoff state, or durable project knowledge would help a coding task.

Use when the task depends on commit history, what changed, when a fix shipped, or linking engineering evidence to reasoning memory.

Use when durable project knowledge, gotchas, workflows, or repeated fixes should become reusable agent guidance instead of ordinary memory.

Use when a main agent needs Memorix to coordinate explicit subagent work through tasks, handoffs, messages, file locks, or the orchestrate CLI.

Use when a technical decision, trade-off, rejected alternative, architecture rationale, or design risk should be recorded or recovered.

Use when resuming work, preparing handoff context, binding an HTTP control-plane project, or deciding whether session_start is useful.

Use when choosing or confirming tech stack for a product or feature: frontend framework, database, auth, email, payments, hosting, search, or agent runtime. Applies AVID defaults (React/Vue, SQLite→Postgres/Supabase, Resend, Stripe, Playwright, Tavily,…

Use when starting product, feature, build, stack, research, agent-framework, or implementation work. Loads AVID personal workflow defaults and skill routing. Defaults are advisory and scene-dependent — not absolute hard blocks. Skip or adapt when the…

Use when choosing or implementing agent runtimes: LangGraph workflows, Pi / pi-ai / embedded Pi harness agents, thin provider SDK loops, model API wire protocol (Chat Completions vs Responses vs Messages), or "which agent framework" decisions. Scene-based…

Prefer when a required CLI, library, MCP, or skill is missing. Install official or best community option when practical. Advisory — hand-roll is allowed with an explicit reason (license, abandoned, fundamental mismatch, user order).

Use for high-stakes decisions: architecture forks, build-vs-buy, competitive response, major refactors, or when the user wants multi-agent debate. Spawns diverse perspectives (or Codex second opinion) and synthesizes a recommendation. Do NOT use for trivial…

Prefer before greenfield products, new verticals, or "做个X/从零做" work. Advisory research + grill-me + MVP cut. Strongly recommended to avoid blind builds, but not an absolute block if the user already has research/spec or explicitly wants a spike. Do NOT use…

Prefer before self-implementing a non-trivial library, service, or subsystem. Advisory find-before-found. Not a hard block on prototypes the user accepts.

Use when designing knowledge bases, RAG, wiki, knowledge graphs, memory for agents, or "how should we store docs for humans vs agents". Produces a clear consumer/form/retrieval decision. Avoids buzzword soup. Do NOT use for ordinary app CRUD databases.