$ loading_
Turn Claude Code and Codex transcripts into interactive local run graphs.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "rungraph" yet — see the docs or source repo.
Use rungraph to list the available runs and show basic information for each run in chronological order.
A list of readable runs so the user can choose which transcript to inspect.
Get the graph for a selected run, then find nodes related to errors, retries, or failures, and summarize their locations and relationships.
The run graph plus located error-related nodes and a brief analysis of how they connect.
Open the visualization for this run, focus on a specified set of nodes, and report the current focused view.
An interactive visualization is opened, and the current focused node range or view state is reported.
After a Claude Code or Codex session, developers can read local transcripts and turn them into a run graph to review steps, branches, and key nodes. This makes the generation or editing process easier to understand.
When an AI agent session is long and contains many nodes, users can use search and detail tools to locate specific nodes and focus on relevant sections. This helps identify anomalies, repetition, or important turning points faster.
For teams that do not want extra runtime hooks or dependencies, rungraph analyzes transcripts after the fact. Its local-only approach and zero runtime dependencies suit lightweight reviews.
It is an MCP server that turns Claude Code and Codex transcripts into an interactive run graph. It provides seven tools: list_runs, get_graph, find_nodes, get_detail, focus_nodes, get_current_view, and open_visualization.
The provided information says it supports stdio and can be run with "npx rungraph mcp". It is published on npm as rungraph.
According to the description, it reads transcripts post-hoc and does not require hooks. It is also local-only and has zero runtime dependencies.
Build a knowledge graph from repos for Q&A and implementation planning.
Connect a local GraphRAG knowledge base for private offline document retrieval.
Query codebase knowledge graphs directly for faster, lower-token code understanding.
Provide structured code context and dependency graphs for efficient codebase understanding.
Explore codebases with semantic search, project graphs, history, and web context.
Build, analyze, and visualize relationship graphs with powerful graph algorithms.