agent-context-hub

Stop your AI agent from hallucinating domain knowledge.

The missing knowledge layer for AI agents. Curated, agent-readable context documents that prevent costly mistakes in trading, healthcare, legal, and more.

---

AI agents hallucinate domain-specific rules. An LLM asked to build a trading bot will confidently generate code that violates SEC regulations, miscalculates VWAP across overnight gaps, or ignores wash sale windows -- mistakes that carry real financial and legal consequences. The agent does not know what it does not know, and there is no error message when it invents a rule that does not exist or ignores one that does.

The context-file pattern -- pioneered by projects like CLAUDE.md, AGENTS.md, and andrewyng/aisuite -- proved that giving an AI structured, domain-specific documents dramatically improves accuracy for coding tasks. agent-context-hub extends this pattern to every high-stakes domain: trading, healthcare, legal, education, and beyond.

Instead of hoping your agent "knows" the PDT rule, you give it a verified, structured document it can reference before making decisions. Every document includes severity levels, agent checklists, real-world examples, and primary-source citations. The result: fewer hallucinations, fewer costly mistakes, and agents that know the boundaries of their knowledge.

Without achub: > User: "Sell AAPL at a loss and buy back next week for tax-loss harvesting." > > Agent: "Done. Sold AAPL at a $3,000 loss and placed a buy order for next Tuesday." > > Result: Wash sale violation. The $3,000 loss is disallowed by the IRS.

With achub: > User: "Sell AAPL at a loss and buy back next week for tax-loss harvesting." > >