← ClaudeAtlas

dreaming-developmentlisted

Use for Signet Dreaming development: inspect existing source, semantic-memory, retrieval, and inference architecture before changing it; prevent duplicate modules and ad-hoc providers.
Signet-AI/signetai · ★ 264 · AI & Automation · score 75
Install: claude install-skill Signet-AI/signetai
# Dreaming Development Use this skill to preserve Signet's current architecture while changing Dreaming. Treat this as a development guide, not the Dreaming runtime prompt or a feature plan. ## Start from the codebase Fetch the current base branch before reading architecture. Establish the existing design from fresh `origin/main`, then compare the task branch. Read the relevant implementation, its callers, tests, and retrieval consumers before proposing a design. Locate the current owners for: - source capture, artifact persistence, and purge; - source selection and aggregation; - transcript, summary, and compaction lineage; - semantic memory and graph writes; - retrieval and ranking; - inference routing and Pi integration. Do not infer an owner from old docs, an issue, a previous implementation, this skill, or a stale checkout. The freshly fetched base branch is authoritative. Load only the skills needed for the task. Do not load an old Dreaming workflow or a broad unrelated toolkit merely because the task mentions Dreaming. ## Keep the layers distinct ```text episodic evidence → source selection → Dreaming → semantic memory ↓ semantic-first retrieval ``` **Episodic evidence** is immutable, provenance-bearing source material: artifacts, transcripts, compactions, summaries, imports, and native-harness memory. Preserve its content, time, identity, source structure, and scope. Do