arch_context
Read-only architectural memory. Returns components, decisions, and lessons relevant to the query with updated N ago stamps and a minscore cutoff.
How to use it
arch_context is exposed by the Paparats MCP MCP server. Add the server to your MCP client (Claude Desktop, Cursor, Windsurf and others), and the arch_context tool becomes available to the model automatically. See the full listing for setup details and every tool this server provides.
Install Paparats MCP
claude mcp add --transport http paparats http://localhost:9876/mcpOther tools in Paparats MCP (20)
Record a component with Does / Owns / Does not / Touched when fields. Idempotent by name.
Record an ADR-style decision (context / decision / alternativesrejected / consequences). Server-side similarity gate refuses duplicates and surfaces near-matches; supersedes links replace prior decisions.
Record a lesson as rule / why / when. Duplicates bump updatedAt (Reflexion confirmation) instead of overwriting.
Off-anchor result share per user — indicator of search noise.
Wipe Qdrant chunks + SQLite metadata for a project (CLI's paparats remove calls it).
Comprehensive feature analysis: locations + recent changes for a question.
AST parse failures, regex fallbacks, zero-chunk files, binary skips.
Walk the symbol graph from a chunkid — incoming (callers/references in), outgoing (calls/references out), or both.
Retrieve a chunk by ID with optional surrounding context.
Git history and ticket references for a chunk — commits, authors, dates. No code.
Indexing status, chunks per group, running jobs.
Cross-chunk impact for a chunkid — symbol graph traversal + cross-project blast radius.
List indexed projects with chunk counts and detected languages.
Timeline grouped by date with commits, tickets, affected files. since filter.
Tool-call retry rate per user — indicator of unhelpful results.
Semantic search filtered by last-commit date. Each result shows when it last changed.
Semantic search across indexed projects. Returns chunks with symbol info and confidence scores.
Top-N slowest searches with timing + chunk counts.
Aggregate token-savings stats (naive baseline vs search-only vs actually consumed).
Most frequent queries by user/session/project anchor.