Run a daily AI-driven strategy review that turns activity logs into concrete actions instead of a long essay.
Track situation, main bets, pending decisions, and red lines in five rolling strategy files updated every review.
Escalate decisions that have stalled for days into a short digest that only needs a one-word reply.
Write a custom doctrine pack (such as Stoic philosophy or Boyd's OODA loop) to change the lens the tool uses.
| ai-nate/miaosuan | 0xblackash/cve-2026-46242 | 1038lab/comfyui-agnes-ai | |
|---|---|---|---|
| Stars | 18 | 18 | 18 |
| Language | — | C | Python |
| Setup difficulty | easy | easy | easy |
| Complexity | 2/5 | 1/5 | 2/5 |
| Audience | pm founder | ops devops | developer |
Figures from each repo's GitHub metadata at analysis time.
No API keys needed, installs as a skill file read by an existing agent like Claude Code.
Miaosuan is an open-source system that gives AI agents a daily strategy-review loop, meant to turn scattered decisions into a single action instead of a long essay. Its name comes from a line in Sun Tzu's Art of War about calculating carefully before a battle. Rather than being another prompt pack that asks an AI to analyze a business through an old military text, miaosuan is a working system: it reads what actually happened each day, interprets it through a pluggable strategic doctrine, forces every insight into an action, and compresses stalled decisions into a short decision card the user can answer in one word. The loop runs in five steps: gather activity from the review window, interpret it through doctrine lenses, force each interpretation into an action, update a set of five rolling strategy files, and escalate any decision that has stalled for several days into a thirty second digest. A human reply of one word feeds back into the next cycle. The Seven Military Classics, a set of ancient Chinese strategy texts, ship as the first doctrine pack, distilled into thirty five operational lenses, alongside full per chapter study notes for all seven texts. The five rolling strategy files track the situation on each front, the one to three main bets everything else serves, pending decisions waiting on a human with age counters, hard red lines the agent must never recommend, and a dated changelog of every strategy change. Doctrine packs are pluggable, so other frameworks like Stoic philosophy or Boyd's OODA loop could be added as separate packs following a documented spec. Setup works by cloning the repository, installing SKILL.md as a skill for an agent such as Claude Code or OpenClaw, copying the strategy templates into the agent's workspace, and running the review manually or on a daily schedule. No API keys or accounts are required since it runs entirely as markdown read by an existing agent. Everything stays local to the user's workspace and nothing is sent elsewhere. The skill and code are MIT licensed, while the doctrine packs are released under CC-BY-4.0.
Miaosuan is an open-source daily strategy-review tool that has an AI agent read what happened each day, interpret it through pluggable doctrine lenses, and compress stalled decisions into a one-word decision card.
Skill and code are MIT licensed (free to use and modify), doctrine packs are CC-BY-4.0 (free to use with attribution).
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly pm founder.
This repo across BitVibe Labs
Verify against the repo before relying on details.