Indexes Architecture Decision Records and extracts dependency graphs into AgentDB memory for rapid semantic searching and tracking.
The ADR Indexer parses and indexes Architecture Decision Records (ADRs) across projects, automatically converting on-disk markdown documentation into structured memory namespaces and relationship graphs. It supports multiple ADR formats (v3 heading styles and YAML frontmatter) and maps dependency edges such as supersedes, amends, and related-to without redundant tool round-trips. This skill is essential for bootstrapping architecture knowledge in a new codebase, syncing ADR graphs after imports, and enabling fast semantic searches over historical engineering decisions.
Key Features
01Upserts records to memory namespaces with deduplication and change detection
02Filters commit hashes and PR numbers to avoid false-positive ADR reference IDs
03Enables high-speed semantic search across historical architectural context
04Parses both legacy heading-style and YAML frontmatter ADR formats in bulk
0569,650 GitHub stars
06Extracts and registers semantic relationship edges (supersedes, amends, depends-on)
Use Cases
01Enabling semantic querying of past technical architecture choices and rationale
02Bootstrapping and organizing architectural decision tracking on existing repositories
03Synchronizing agent memory graph with updated or imported on-disk ADR files