Skip to content

v0.2.0 — Live Dashboard, Documentation & Reliability

Choose a tag to compare

View all tags
@ssdavidai ssdavidai released this 23 Feb 14:49
v0.2.0
cc2092f

What's New

Live Dashboard (alfred up --live)

  • Per-worker interpreted feeds — 2x2 grid showing each worker's pipeline progress in real time
  • 60+ structlog events mapped to human-readable messages (e.g. "Stage 1 done — 3 entities found")
  • Silent failure detection — flags anomalous "successes" like 0 entities created or empty runs
  • Health indicators — healthy/degraded/failing/stopped/restarting per worker
  • Adaptive layout — 1-4 workers automatically arranged

Reliability Improvements

  • Manifest retry loops — curator and distiller now retry up to 3x when LLM fails to write manifest
  • Entity relevance filtering — curator uses user profile context to avoid creating records for TV shows, celebrities, and other irrelevant entities
  • Base-view preservation — vault records created with custom body content now retain their Dataview template sections

User Profile

  • Generic user-profile.md template created during quickstart
  • Curator uses it to scope entity extraction to what's relevant to you
  • Searched at {vault}/user-profile.md then ~/.config/alfred/user-profile.md

Documentation

  • 13 documentation pages covering all 4 workers, CLI commands, configuration, vault schema, live dashboard, architecture, agent backends, user profile, and installation
  • Updated README with full feature documentation

Bug Fixes

  • Fixed Ollama 500 errors by reducing embedding text limit
  • Added staggered worker startup to reduce Ollama load
  • Fixed OpenClaw session deadlocks with separate agents per tool
  • Fixed YAML crash in janitor autofix
  • Fixed OS arg length limit in OpenClaw backend
  • Added crash protection to all daemon loops

Install / Upgrade

pip install alfred-vault==0.2.0

Full docs: https://github.com/ssdavidai/alfred/tree/master/docs

Assets 4
Loading