aleksandr-bogdanov
UserPlain-markdown memory for AI assistants. The model writes once, grep + BM25 reads it back. No embeddings, no MCP server. MIT.
Categories
Indexed Skills (3)
imprnt-release
Cut a stable imprnt release to npm `latest`. Detects affected packages since the last release, patch-bumps each, and fires the gated release workflow. Use when the user says "imprnt-release", "/imprnt-release", "ship imprnt", "release imprnt", "promote imprnt to latest". DEFAULT IS NO PROSE, just ship.
imprnt-hot
Prime a fresh session from the imprnt vault — surface the needs-review backlog (ingest soft-fails, unresolved entities) plus the ~500-token current-context primer. Explicit-on-request, never auto-injected. USE WHEN what's active, prime me, what needs review, catch me up, hot, session primer, what was I doing, imprnt hot.
imprnt-recall
Load SCOPED context from the imprnt vault by topic. Turn the user's question into keywords + candidate tags, run the deterministic ranked recall, read only the top hits, and present the relevant notes — never the whole vault, never auto-injected. USE WHEN load context, load my context about, what do I know about, pull context on, recall, brief me on, imprnt recall.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.