@djolex999/vir-cli 0.6.1 → 0.6.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (2) hide show
  1. package/README.md +75 -40
  2. package/package.json +1 -1
package/README.md CHANGED
@@ -18,7 +18,7 @@ developer-tools, mcp, local-first, cross-platform, llm-wiki
18
18
  <a href="https://www.npmjs.com/package/@djolex999/vir-cli"><img src="https://img.shields.io/npm/v/@djolex999/vir-cli?color=7c6af7&label=npm" alt="npm version"></a>
19
19
  <a href="https://www.npmjs.com/package/@djolex999/vir-cli"><img src="https://img.shields.io/npm/dw/@djolex999/vir-cli?color=4fd1a0" alt="npm downloads"></a>
20
20
  <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-22d3ee" alt="license"></a>
21
- <a href="#project-status"><img src="https://img.shields.io/badge/tests-79%20passing-22c55e" alt="tests"></a>
21
+ <a href="#project-status"><img src="https://img.shields.io/badge/tests-85%20passing-22c55e" alt="tests"></a>
22
22
  <a href="#project-status"><img src="https://img.shields.io/badge/platforms-macOS%20%7C%20Linux-lightgrey" alt="platforms"></a>
23
23
  <a href="https://modelcontextprotocol.io"><img src="https://img.shields.io/badge/MCP-server-c084fc" alt="mcp"></a>
24
24
  <a href="#"><img src="https://img.shields.io/badge/local--first-yes-f59e0b" alt="local-first"></a>
@@ -27,18 +27,25 @@ developer-tools, mcp, local-first, cross-platform, llm-wiki
27
27
 
28
28
  ## The pattern
29
29
 
30
- Recently, Andrej Karpathy described a pattern he calls the **LLM Wiki** — AI work
31
- that feeds back into itself through a persistent, curated, structured artifact,
32
- instead of resetting at the end of every session. He ended his post with: _"I
33
- think there is room here for an incredible new product instead of a hacky
34
- collection of scripts."_
35
-
36
- Vir is one implementation of that pattern, with Obsidian as the frontend.
37
-
38
- > Vir reads two input sources today: Claude Code session transcripts (`.jsonl`)
39
- > and web articles (markdown clipped via Obsidian Web Clipper). Both get
40
- > distilled into the same vault. Future versions will add PDFs, code repos, and
41
- > images — matching the full LLM Wiki pattern.
30
+ In April 2026, Andrej Karpathy described a pattern he calls the **LLM Wiki** — AI
31
+ work that feeds back into itself through a persistent, curated, structured
32
+ artifact, instead of resetting at the end of every session. He published the idea
33
+ file at [karpathy/llm-wiki.md](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f)
34
+ and ended his
35
+ post saying: _"I think there is room here for an incredible new product instead of
36
+ a hacky collection of scripts."_
37
+
38
+ Several open source implementations of this pattern now exist
39
+ ([lucasastorian/llmwiki](https://github.com/lucasastorian/llmwiki),
40
+ [Pratiyush/llm-wiki](https://github.com/Pratiyush/llm-wiki),
41
+ [nashsu/llm_wiki](https://github.com/nashsu/llm_wiki) among them). Each takes a
42
+ different shape.
43
+
44
+ Vir is the Obsidian-native one. It treats Obsidian as the primary frontend — not
45
+ just a storage location — and integrates deeply (sidebar plugin coming,
46
+ dataview-compatible frontmatter, canvas integration planned). It reads AI coding
47
+ session transcripts retroactively, so months of existing history become a
48
+ queryable knowledge base in one run.
42
49
 
43
50
  [Karpathy's post →](https://x.com/karpathy/status/2039805659525644595)
44
51
 
@@ -57,6 +64,20 @@ implementation of [the pattern above](#the-pattern).
57
64
  <img src="assets/demo.gif" width="800" alt="vir distilling Claude Code sessions into notes in an Obsidian vault">
58
65
  </p>
59
66
 
67
+ ## What's coming
68
+
69
+ Vir is actively developed. In the next 30–60 days:
70
+
71
+ - **Obsidian plugin v1** — sidebar, command palette, canvas integration,
72
+ submitted to the community plugin marketplace
73
+ - **Multi-agent support** — Codex CLI, Cursor, Aider, Cline (one per release)
74
+ - **Topic synthesis** — `vir compose` generates LLM Wiki-style topic pages that
75
+ merge insights across sessions
76
+ - **PDF/paper ingestion** — broaden beyond developer workflows
77
+
78
+ Track progress via [GitHub issues](https://github.com/djolex999/vir/issues) or
79
+ watch the repo for releases.
80
+
60
81
  ## Quality controls
61
82
 
62
83
  Auto-distilled notes can be wrong. The most common concern from early users:
@@ -125,32 +146,41 @@ better sessions
125
146
 
126
147
  ## How Vir compares
127
148
 
128
- The AI memory space has grown fast. An honest comparison to the options worth
129
- knowing about:
130
-
131
- | | Vir | claude-mem | claude-memory | mem0 |
132
- | ----------------------------------- | ------------------------------- | ----------------------- | -------------------------- | ----------------- |
133
- | Reads existing Claude Code sessions | ✓ | from install forward | from install forward | n/a |
134
- | Markdown output (Obsidian-native) | ✓ | ChromaDB | LanceDB | various backends |
135
- | MCP server | ✓ | ✓ | ✓ | n/a |
136
- | Setup complexity | `npm install -g` | Bun + Python + ChromaDB | pnpm + LM Studio + LanceDB | API/cloud setup |
137
- | Cross-platform daemon | mac launchd, linux systemd/cron | mac, linux | mac, linux | n/a |
138
- | Open source license | MIT | Apache 2.0 | MIT | open core + cloud |
139
-
140
- **Different tools for different needs:**
141
-
142
- - Want a heavyweight memory plugin with real-time capture and vector storage?
143
- Use **claude-mem**.
144
- - Want sophisticated retrieval (MMR diversity, web dashboard, multi-phase
145
- maintenance)? Use **claude-memory**.
146
- - Building AI applications that need to remember users long-term? Use **mem0** —
147
- it's infrastructure for apps.
148
- - Want your Claude Code sessions distilled into markdown notes you can browse,
149
- edit, and own in Obsidian? Use **Vir**.
150
-
151
- These aren't all competitors. mem0 is a different layer of the stack entirely.
152
- claude-mem and claude-memory share the same input data as Vir but take different
153
- opinions on storage and integration.
149
+ The LLM Wiki space has grown fast. Honest comparison:
150
+
151
+ ### vs other LLM Wiki implementations
152
+
153
+ | | Vir | lucasastorian/llmwiki | Pratiyush/llm-wiki | nashsu/llm_wiki |
154
+ | --------------------------------- | ---------------------------- | ----------------------------- | ---------------------------------------- | ------------------------- |
155
+ | Language | TypeScript / Node | Python | Python | Cross-platform desktop |
156
+ | Distribution | `npm install -g` | Local app + hosted SaaS | `git clone` + python | Desktop app installer |
157
+ | Obsidian integration | Native (plugin v1 in dev) | Markdown output | Outputs to vault | Own UI, no Obsidian |
158
+ | Input sources | Claude Code, Web Clipper (more agents coming) | PDFs, docs upload | Claude Code, Cursor, Cline, Codex, Gemini | Documents, mixed sources |
159
+ | Retroactive on existing sessions | ✓ | n/a | from install forward | n/a |
160
+ | MCP server | ✓ | ✓ | ✓ | ✓ |
161
+ | License | MIT | open source + hosted commercial | MIT | open source |
162
+
163
+ ### vs Claude Code memory tools
164
+
165
+ | | Vir | claude-mem | claude-memory | mem0 |
166
+ | ----------------------------------- | ---------------- | -------------------- | ------------------ | ----------------- |
167
+ | Reads existing Claude Code sessions | ✓ | from install forward | from install forward | n/a |
168
+ | Markdown output | ✓ | ChromaDB | LanceDB | various backends |
169
+ | Setup | `npm install -g` | Bun + uv + Python | pnpm + LM Studio | API/cloud setup |
170
+ | License | MIT | Apache 2.0 | MIT | open core + cloud |
171
+
172
+ ### Different tools for different needs:
173
+
174
+ - **If you want a polished cross-platform desktop app** for general document
175
+ knowledge bases, use lucasastorian/llmwiki or nashsu/llm_wiki.
176
+ - **If you want multi-agent support** with rich entity/concept page taxonomy and
177
+ don't care about Obsidian integration depth, use Pratiyush/llm-wiki.
178
+ - **If you want a heavyweight Claude Code memory plugin** with real-time capture
179
+ and vector storage, use claude-mem.
180
+ - **If you're building AI applications that need to remember users** long-term,
181
+ use mem0 (different layer entirely).
182
+ - **If you want an Obsidian-native LLM Wiki** that reads your existing Claude Code
183
+ sessions and is on its way to supporting multiple AI coding agents, use Vir.
154
184
 
155
185
  ## Real-world results
156
186
 
@@ -165,6 +195,11 @@ Real output from the author's first run across 226 Claude Code sessions.
165
195
  | Projects covered | 8 projects |
166
196
  | Knowledge breakdown | 54 patterns · 47 decisions · 23 gotchas · 2 tools |
167
197
 
198
+ These numbers show Vir works at scale, not that it's uniquely capable. Other LLM
199
+ Wiki implementations would produce similar results with the same input. The
200
+ distinguishing question for Vir is workflow fit — does the Obsidian-native +
201
+ retroactive + multi-agent positioning match how you actually work?
202
+
168
203
  Example query against the distilled vault:
169
204
 
170
205
  ```bash
@@ -382,7 +417,7 @@ vault/vir/
382
417
 
383
418
  | | |
384
419
  | -------------- | ----------------------------------------- |
385
- | Tests | 79 passing |
420
+ | Tests | 85 passing |
386
421
  | Platforms | macOS (launchd), Linux (systemd/cron) |
387
422
  | Node | 20+ |
388
423
  | First-run cost | $1–5 (Kie.ai recommended for 72% savings) |
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@djolex999/vir-cli",
3
- "version": "0.6.1",
3
+ "version": "0.6.2",
4
4
  "description": "An LLM Wiki for Claude Code, in your Obsidian vault.",
5
5
  "author": "Djordje Marković <djordje@growthq.rs>",
6
6
  "license": "MIT",