Consolidate runtime artifacts into redactable learnings with routed dispositions.
npx skills add https://github.com/quanpersie2001/pulse --skill dream-quanpersie2001
Please help me install this Agent Skill. Skill: dream Source: https://github.com/quanpersie2001/pulse/tree/main/plugins/pulse/skills/dream Command: npx skills add https://github.com/quanpersie2001/pulse --skill dream-quanpersie2001
This Skill consolidates transient runtime artifacts and scattered agent outputs into durable, redactable learnings so teams have a reviewed, auditable knowledge base and avoid silent drift or unsafe rewrites.
Run a dream consolidation pass over the last 7 days of runtime artifacts and produce merge, correction, ratchet, or pending recommendations.
High-intent search queries and answers about installing and using this skill.
You can consolidate runtime artifacts into durable memory by running a consolidation pass that extracts signals, redacts PII, normalizes dates, and routes outputs into merge, correction, ratchet, or pending dispositions for an auditable knowledge base.
Redact secrets and PII from runtime learnings by treating all agent artifacts as untrusted input, enforcing mandatory redaction and date normalization during signal extraction before routing into durable memory.
Signal routing works by extracting candidates from runtime evidence, classifying them as patterns or corrections, resolving contradictions, and assigning each an exact disposition like merge, create, ratchet, or skip to ensure safe knowledge base updates.
Yes, you can use this to bootstrap memory from mixed Claude and Codex runtime evidence. It detects bootstrap versus recurring runs, writes run-level provenance, and performs targeted consolidation while preserving operator control.
The limitation of automatic critical-pattern promotions is that they are blocked; the system never edits critical-patterns without explicit approval, enforces exact-one-owner rewrite rules, and leaves ambiguous items pending behind explicit operator choices.
supertyrelleCombine pipe-format result blocks into one deduplicated unified output.
netbrainMonitor Ratchet PRs for merge conflicts, CI failures, and review comments.
minhtri2710Consolidate codex-derived learnings into .beads/learnings with provenance checks.
AissenLiuAutomate GPU-accelerated data curation workflows for large multimodal datasets.
