Browse / Second brain / LLM wiki / sandbox
sandbox
steve-lemon
sandbox for second brain with obsidian wiki
Open steve-lemon/sandbox → Checked against the four rules.
Works with any agent
ships AGENTS.md, the cross-tool instruction file.
In its own words
Sandbox for testing Knowledge Ops Setup by Steve.
Personal knowledge base built with Claude Code — raw sources compiled into structured wiki articles via an automated Reader → Writer pipeline.
Usage
This sandbox is the initial environment for a future production knowledge folder. Treat the repository root as VAULT_DIR only when the expected vault structure is present (VAULT_RULES.md, wiki/, raw/, Clippings/, outputs/, and templates/). Do not silently fall back to ~/knowledge.
Drop markdown files into Clippings/ and run the ingest pipeline:
Automation should prefer Claude Code for ingest when the claude CLI is installed and authenticated. If Claude Code is unavailable, the agent falls back to the Hermes-native vault-ingest workflow. The pipeline moves files to raw/, extracts concepts, and creates or updates wiki/ articles with frontmatter, wikilinks, and topic index entries.
For explicit delegation, ask:
Lint follows the same policy: prefer Claude Code for scheduled/delegated lint, then fall back to Hermes-native lint if Claude is unavailable.
Query
Ask any question — the agent reads wiki/INDEX.md, identifies relevant articles, and writes a cited answer to outputs/.
The whole README. Open the repo ↗
Read from the repository's git tree on 2026-08-17.