🤓 Latest Submissions
Repo-Insight : Graph-RAG Knowledge for IBM Bob
Repo-Insight parses any Python repository into a FalkorDB property graph and exposes 22 MCP tools that IBM Bob calls directly during bug-fixing tasks. When a test fails, Bob uses our deterministic stack-trace localizer to pinpoint the exact function responsible — no semantic guessing, no false positives — then narrows the blast radius from dozens of suspects to only the functions that actually executed during the failure, eliminating 83% of irrelevant context. Instead of reading full source files, Bob queries compact function fingerprints that describe each function in ~30 tokens versus ~500 for raw source, cutting total token consumption by 15× on real bugs in production codebases like Django. The result is an agent that navigates 500,000-line repositories with the precision of a senior engineer who already knows the codebase — finding the right file, reading only what matters, and fixing the bug in a fraction of the cost.
17 May 2026
Repo-Insight : Next-Gen Coding Assistant
Repo-Insight is a graph-driven AI coding agent designed to eliminate cascading codebase breakages caused by standard large language models through an anti-RAG paradigm. Powered by FalkorDB and optimized for high-concurrency inference engines like SGLang, it shifts software engineering from naive text chunking to guaranteed structural completeness. Repo-Insight idempotently parses repositories via Tree-sitter into highly relational abstract syntax tree entities—mapping functions, classes, modules, call edges, imports, and inheritance—and embeds them concurrently. This acts as a strict control plane, enforcing deterministic graph traversals before any code modification occurs. The agent utilizes an autonomous, six-phase execution pipeline called Mode C. It syncs repository modification fingerprints, deploying surgical Cypher queries to clean stale files before localization combines vector search with language model extraction to pinpoint defect entry points along traceback paths. The engine deterministically expands the change subgraph to map the upstream blast radius of components that will break and the downstream impact radius of touched dependencies. A planning phase forces coverage across every exposed file, utilizing an autonomous validation gate to intercept hallucinations and drive self-correction before generating precise search-and-replace edit blocks. Finally, edits are applied to a hard-linked local sandbox, verified against test suites, and instantly re-ingested to compute post-edit graph deltas. Supported by a Streamlit studio, live debouncing watchers, a Model Context Protocol server, and SWE-bench orchestrators, Repo-Insight proves structured graph reasoning yields higher precision and safer codebases than pure parameter scale.
10 May 2026
