Graph Systems
Knowledge-Driven Agent Architectures
Use ontologies and graph evidence to constrain plans without confusing knowledge with authority.
10 chapters · 6-8 hours reading · 4-5 hours labs · 10-15 hours project
Chapters
- 1The knowledge agent as a bounded transition systemMental model
- 2Typed plans, preconditions, and action schemasFormal planning
- 3Ontology-constrained planning and validationSemantic control
- 4Graph evidence acquisition and claim ledgersKnowledge retrieval
- 5Temporal state, memory, and event historyDurable cognition
- 6Tool orchestration, budgets, and durable executionRuntime engineering
- 7Staged writes, authorization, and compensationSafe action
- 8Knowledge-agent threat model and fail-closed designSecurity engineering
- 9Trajectory evaluation, counterfactual replay, and red teamsAgent evaluation
- 10Production governance, observability, and capstoneOperational mastery