Bunker OS is a folder of plain Markdown files that doubles as an OpenCode agent environment. There is no installation wizard, no cloud setup, and no account required — you clone the repository, open it in Obsidian, and point OpenCode at the same directory. The hooks, skills, and scripts are already wired together.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/SamBleed/opencode-obsidian/llms.txt
Use this file to discover all available pages before exploring further.
The recommended way to use Bunker OS is to clone it directly as your Obsidian vault. This gives you the full repository structure, pre-configured hooks, all 13 skills, the test suite, and the n8n automation layer.
opencode-obsidian/ directoryObsidian will load the vault with pre-configured community plugins (Calendar, Thino, Excalidraw, Banners) and CSS snippets already in place. The graph view, dashboard, and canvas layer are all ready to use immediately.
The
wiki/ directory is gitignored — your personal notes, research, and knowledge pages stay local and never leave your machine.With the vault open in Obsidian, open OpenCode in the same
opencode-obsidian/ folder. OpenCode will automatically discover the hooks/hooks.json file and register the four lifecycle hooks.{
"hooks": {
"SessionStart": [ ... ],
"PostCompact": [ ... ],
"PostToolUse": [ ... ],
"Stop": [ ... ]
}
}
At session start, OpenCode reads
wiki/hot.md silently to restore recent context. You can confirm the SessionStart hook is working by checking the hooks file directly:wiki/hot.md existswiki/hot.md to restore context without announcing itThe
Stop hook detects whether wiki/ was modified during the session and, if so, prompts OpenCode to update wiki/hot.md with a summary of changes — so the next session picks up exactly where this one left off.Read
wiki/hot.md at the start of any session where you want to resume previous work. It is a ~500-word summary of the most recent context: current focus, key recent facts, recent changes, and active threads.With OpenCode running inside
opencode-obsidian/, you have access to all 13 bundled skills immediately. Start with any of these:Autonomous research — 3-round web research loop that decomposes a topic into angles, searches and fetches sources, cross-references for gaps, and files everything into the wiki:
Wiki query — reads
wiki/hot.md, then wiki/index.md, drills into relevant pages, and synthesizes an answer from what is actually in the wiki:BM25 text retrieval — keyword search over wiki pages using the pure-Python BM25 index (no embeddings, no LLM needed for indexing):
autoresearch [topic]retrieve [query]think [problem]ingest [file]ingest all of thesewhat do you know about Xlint the wikisave thiscanvascanvas add image [path]make test # 430 tests, 5 suites
./bin/bunker-check.sh # Local definition of done
./bin/wiki-integrity.sh # Scan for orphans and broken links
./bin/evidence-index.sh # Index evidence with SHA256
./bin/bunker.sh init # Vault health check
./bin/wiki-sync.sh --apply # Sync + commit wiki
python3 scripts/retrieve.py build # Rebuild BM25 index
python3 scripts/retrieve.py "query" # BM25 text retrieval
n8n is Bunker OS’s async nervous system — it handles webhooks, AI triage, GitHub issue creation, and multi-channel notifications. It runs entirely on Docker and communicates with OpenCode through an MCP bridge.
To activate the full AOC v4 Enterprise pipeline (37 nodes, AI triage via OpenRouter, GitHub issue creation), configure the environment variables in
automation/n8n-lab/.env with your OpenRouter API key, Discord webhook URL, and GitHub credentials, then import and activate bunker-aoc-v4-enterprise.json from the n8n UI.Option 2: Use as a Reference from Another Project
If you already have an OpenCode project and want to point it at Bunker OS as a shared knowledge base without making it your primary vault, add a reference block to your project’sAGENTS.md:
Next Steps
Once you have Bunker OS running, the highest-leverage first actions are:- Drop a source into
.raw/and runingest [filename]— the skill extracts entities, concepts, and relationships and creates 8–15 cross-referenced wiki pages automatically. - Run
autoresearch [topic]on any domain you work in regularly — it seeds the wiki with well-sourced, cross-referenced pages that all future sessions can build on. - Run
make testonce to confirm your environment is healthy — all 430 tests should pass.