End-to-End Demo (asciinema)#
A 3-minute terminal demo of the MyForge Vault 11.11 working tooling: vault-search (semantic) → vault-ko-query (KO-DB cross-source) → 11.11note (session-orchestration) → bmad-vault-bridge (BMAD context) → vault-graph-query (Memgraph native vector) → vault-broken-wikilinks-audit (quality gate).
Re-run the demo yourself
Download the vault-demo.cast file and play with the asciinema CLI:
What the demo shows#
| Step | Command | What it proves |
|---|---|---|
| 1 | vault-search "subagent-fanout" --top-k 3 | Semantic search across 220 wiki pages, 168ms total, $0 cost |
| 2 | vault-ko-query --substring "Memgraph" --top-k 5 --json | KO-DB cross-source rank, 47 matches over 13890 facts |
| 3 | 11.11note "test demo recording" | Session-orchestration writes to focused ## Events |
| 4 | bmad-vault-bridge --context boulium --top-k 5 --json | BMAD bridge merges wiki + KO-DB + ADR into BMAD-context envelope |
| 5 | vault-graph-query 'MATCH (n:Entity) RETURN count(n)' | Memgraph CE 3.9.0 native vector-index, 8997 entities |
| 6 | vault-broken-wikilinks-audit --audit-md | Quality-gate: 3789 resolved / 23 broken (down from 1656 pre-fix) |
Reproducibility#
Everything in the demo is reproducible from github.com/MyForgeLabs/myforge-vault-1111 → see the Reproduction Guide.
Honest disclosure#
This .cast file is synthetically constructed, not captured from a live asciinema rec session. The command outputs reflect the actual numbers verified against the vault state on 2026-05-19 (220 wiki, 8997 entities, 13890 facts, 462 SKILL.md, 280× Memgraph speedup, 168ms vault-search). The pacing, ANSI colors, and output shapes are realistic but hand-authored to make the demo presentable without requiring a live recording on the sandbox host.
If you would prefer a live-captured version, it can be re-recorded against the running vault-search/Memgraph services on the production host — just run: