Claude Skills by shakedaskayo

> **For agentic workers:** REQUIRED SUB-SKILL: superpowers:subagent-driven-development. `- [ ]` checkboxes. **Goal:** Deliver the dreaming playbook as the `pensieve-dreaming` skill to the spawned Claude CLI via the filesystem (`<cwd>/.claude/skills/`), replacing the hardcoded `dreaming_prompt()`. No `claude_cli.rs` change (uses the existing `cwd` param). Agent-agnostic; safe fallback. **Spec:** `docs/superpowers/specs/2026-06-09-dreaming-via-skill-design.md`. **Co-working note:** the user is ...

Deploy pensieve to production (AWS Fargate + S3 + Supabase) or run a Supabase-backed local test drive. Use when the user asks to deploy pensieve, self-host pensieve in production, set up pensieve on AWS/Supabase, or tear a pensieve deployment down. Drives the `pensieve deploy` CLI wizard (Terraform or Pulumi under the hood).

Recall durable context (prior decisions, preferences, conventions, learnings) and query the user's live data + knowledge graph; persist new durable memories. Recall BEFORE answering anything that may depend on past context or the user's systems; save AFTER the user states something worth keeping long-term. Backed by the Pensieve context engine via the bundled `pensieve` MCP server (memory + data + graph tools).