skills/astro-han__karpathy-llm-wiki
1
0
Fork 0
mirror of https://github.com/Astro-Han/karpathy-llm-wiki.git synced 2026-09-14 19:09:17 +08:00
Code Issues Packages Projects Releases Wiki Activity

docs: bring examples and install instructions up to current protocol

Browse Source 
log-sample.md now shows the real entry types (ingest with Disposition
and Raw, no-material with full path, archived query, lint) instead of
the retired Compile/Update vocabulary that referenced directories this
skill does not use. Manual install instructions include scripts/ so
the evidence checker actually ships.
This commit is contained in:
Yuhan Lei
2026-07-23 22:47:45 +08:00
parent 19b342b981
commit f8e592f979
3 changed files with 19 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+1 -1
View File
@@ -105,7 +105,7 @@ This skill follows the [agentskills.io](https://agentskills.io) open standard:
| Cursor | `npx add-skill Astro-Han/karpathy-llm-wiki` |
| Codex CLI | Copy to `.agents/skills/karpathy-llm-wiki/` |
| OpenCode | `npx add-skill Astro-Han/karpathy-llm-wiki` |
| Other tools | Copy `SKILL.md` and `references/` into the tool's skill directory |
| Other tools | Copy `SKILL.md`, `references/`, and `scripts/` into the tool's skill directory |
## FAQ
examples/README.md
+4 -5
View File
@@ -9,7 +9,7 @@ This directory contains real files from a knowledge base maintained with `karpat
| `claude-code-statusline-landscape.md` | Compiled wiki article with structured data (tables, citations, cross-references) |
| `2026-03-19-claude-code-statusline-landscape.md` | Raw source material before compilation |
| `ai-coding-tools-index.md` | Topic index with one-line summaries |
| `log-sample.md` | Sample entries from operation log |
| `log-sample.md` | Sample entries from operation log (all current entry types) |
## Raw vs Compiled Comparison
@@ -27,9 +27,8 @@ This directory contains real files from a knowledge base maintained with `karpat
## Operation Log
The log records every action:
- `Compile` — new article from source
- `Update` — cascade updates across related articles
- `Lint` — quality checks
- `Query` — archived query results
- `ingest` — source collected into raw/ and compiled (with Disposition: New / Update / Disputed, or `no material` when nothing was worth compiling)
- `query` — archived query results
- `lint` — quality checks and their outcomes
Recent activity shows daily maintenance: 87 entries in the last 7 days alone.
examples/log-sample.md
+14 -23
View File
@@ -1,29 +1,20 @@
# Wiki Log
<!-- Format: ## [YYYY-MM-DD] Action | Topic. Cascading updates use - Updated: sub-items -->
<!-- Format: ## [YYYY-MM-DD] <action> | <subject>. Actions: ingest / query / lint. Ingest entries record Disposition and Raw; cascade updates use - Updated: sub-items. Append-only. -->
## [2026-04-12] Update | content-strategy audience correction
- Target audience corrected from "IT/managers" to "personal knowledge workers" (ops/admin/assistant/marketing)
- Content perspective shifted from "deployer" to "user": not caring about installation, caring about time saved
- Second draft rewrite: from tool selection to specific efficiency scenarios (WeChat group → Lark table 2h → 10min)
- Updated: build-in-public-content-framework.md (audience section + ratio example + answer asset table), _index.md, draft files
## [2026-04-12] ingest | Build in Public content framework
- Disposition: New
- Raw: raw/content-strategy/2026-04-12-x-buildinpublic-thread.md
- Core finding: altruistic content earns distribution, self-narrative does not
## [2026-04-12] Compile | content-strategy (new topic)
- Created articles/content-strategy/ topic, containing _index.md + build-in-public-content-framework.md
- Source: X account @AstroHanRay empirical data + wiki 7 articles cross-application (solo-developer-skills-2026, growth, karpathy-viral-product-analysis, traffic-channel-efficiency, geo-fundamentals, product-sense-and-restraint, opensource-tool-growth-patterns)
- Core finding: altruistic content has distribution, self-narrative doesn't; Xiaohongshu is gstack Build in Public main battlefield
- Action: deleted X pinned tweet (cat food thread, 0 engagement)
## [2026-04-15] ingest | AI coding tools landscape 2026 Q1
- Disposition: Update; Disputed
- Raw: raw/ai-coding-tools/2026-04-15-vendor-benchmark-post.md
- Updated: Claude Code statusline landscape (vendor's throughput claim contradicted the earlier community measurement; marked Status: Disputed)
## [2026-04-12] Update | Chunke GEO plan (Lark DiYndHeofovBUoxebYNcyd62nRf) supplemented based on complete conference minutes
- Scenario-type queries: 6 items upgraded from single-dimension to compound scenario descriptions ("just adopted 3-month kitten sensitive stomach...")
- Added attribution mechanism callout: UTM parameters + Tmall source survey + AI citation URL click tracking
- Updated annotation: noted 2026-04-12 based on 12 guest minutes supplement
## [2026-04-18] ingest | no material: raw/ai-coding-tools/2026-04-18-weekly-trending-recap.md
- Disposition: No material
## [2026-04-12] Update | Yangzi GEO plan v1 supplemented based on complete conference minutes
- Source: articles/geo/ compiled conclusions + raw/geo/2026-04-12 newly added 9 guests
- Output: output/2026-04-08-yangzi-geo-plan-v1.md four changes
- Board proposal "opportunity" section: added "heavy-decision industries most suitable for priority GEO布局" + 8x growth data
- "How to judge success" section: added attribution mechanism (UTM/consultation source/400 phone同比), start after trial success
- B.3 scenario adaptation 6 questions: upgraded to compound scenario descriptions (multi-constraint叠加), matching AI search user intent expression
- C.1.1 content structure: added scenario-based content requirements (restore scenario → selection logic → product line落地)
- Appendix F success tier: attribution mechanism前置为 long-term budget application necessary condition
## [2026-04-20] query | Archived: Grok vs Claude coding comparison
## [2026-04-21] lint | 6 issues found, 4 auto-fixed