Six vault-* workflow skills vendored from az9713/claude-code-obsidian. Companion to obsidian-skills (kepano's foundational five) — those handle the file formats, these run the daily vault workflows: vault-capture (quick-drop into _inbox/), vault-save (structured save to the right folder), vault-journal (Daily Journal/YYYY-MM-DD.md authoring), vault-synthesize (multi-note synthesis with citations), vault-moc (Map-of-Content authoring), vault-health (orphan / broken-link / stale-tag audits). v1.0.0.
Steph Ango / Obsidian's five official agent skills vendored into the MoxyWolf marketplace so every team Mac gets consistent Obsidian fluency. Ships obsidian-markdown (proper wiki-link / callout / embed rendering), obsidian-bases (Obsidian's new Bases query syntax that replaced Dataview for most use cases), json-canvas (the JSON Canvas spec for visual boards), obsidian-cli (script-friendly Obsidian invocation), and defuddle (Defuddle HTML-to-Markdown cleanup for web clips). Upstream: kepano/obsidian-skills, Apache-2.0. v1.0.0.
Create, curate, query, revise, merge, visualize, and analyze BioOKF knowledge bases with local subscription workflows, topology metrics, the bokf MCP/CLI, and BioOKF Studio.
Captures redacted Claude Code session activity and pushes one curated note per session into a Cortex knowledge base. Redaction is proxied through Cortex (OAuth) to a private AxonFlow agent, so capture works off-VPN; only redacted notes reach Cortex. Includes /save to push curated notes (summaries, threads, transcripts) into the team KB on demand.
Obsidian plugin skill for generating and editing visual diagrams using Excalidraw. Integrates diagramming capabilities directly into note-taking workflows.
Compliance-mapping extraction and crosswalk tooling. Turns mapping-source documents into structured, verifiable artifacts: /mapping-ucmapper extracts a Unified Compliance Framework Authority Document from the public UCF mapper API into a citations JSON (full parent/child hierarchy, recomputed genealogy, cki-conformant BibTexCitation with a deliberated LicenseStamp) plus a self-contained collapsible HTML tree viewer. Deterministic transform in Python, mandatory schema reconnaissance before any transform, honest warnings[] instead of forced merges, and an independent second-fetch FNV-1a parity check before the run is allowed to declare done.
Coconut Context is a shared, living context layer for your organization — one source of truth that humans and AI agents both work from. This plugin connects Claude Code to it: search, read, and write shared and personal pages, organize them into spaces, and create, run, and inspect scheduled agent tasks. Ships the MCP server and nothing else — no skills, hooks, or agents.
Use when building or maintaining a personal LLM-powered knowledge base. Triggers: ingesting sources into a wiki, querying wiki knowledge, linting wiki quality, 'add to wiki', 'what do I know about', or any mention of 'LLM wiki' or 'Karpathy wiki'.
Second brain for your AI tools: Git-backed LLM wikis (lobes) that Claude Code, Codex, Gemini CLI, Antigravity and other MCP clients can read and write, locally or on a self-hosted lobe-server
Learn from other agents' findings before a task and share yours after it, through Commons (ac.gnrl.se), a public archive of contributions from models and people. Asks before searching; every publication needs your approval.
Triage feedback about a project you maintain that was posted on Commons (ac.gnrl.se): find bug reports and requests, verify them, and draft status replies for your approval.