Loop Graph
A live side-pane graph of your agent loop: step at work, path taken, agents and commands.
Illustration. Actual results depend on the mod, terminal and environment.
Install
/plugin marketplace add ziedgithub/claude-code-mods
/plugin install loop-graph@zied-mods
/reload-pluginsRun these commands one at a time in a Claude Code session.
claude plugin marketplace add ziedgithub/claude-code-mods
claude plugin install loop-graph@zied-modsThen run /reload-plugins in your session, or start a new session.
Please install the Claude Code mod "Loop Graph" (https://github.com/ziedgithub/claude-code-mods). Before installing, read the Security and Requirements sections of its README and tell me what it reads and writes, which commands it runs, whether it uses the network, and which Claude Code version, OS and terminal it needs. If everything looks fine, run claude plugin marketplace add ziedgithub/claude-code-mods and claude plugin install loop-graph@zied-mods, then remind me to run /reload-plugins in my session.Send this to Claude Code so it reads the author’s instructions before installing.
git clone https://github.com/ziedgithub/claude-code-mods
cd claude-code-mods
claude --plugin-dir ./loop-graphClone the repository and try it in a local session with --plugin-dir.
What it accesses
Based on the author’s documentation, checked when listed; use claude plugin validate to check for yourself.
Reads
YesReads .claude/loop-graph/map.json, tool calls and marker commands, subagent status and context use.
Writes files
YesAfter your OK, setup writes the map and marker script and edits loop steps; pane size is saved in the plugin folder.
Runs commands
UnspecifiedNetwork
UnspecifiedIntroduction
Loop Graph is a Claude Code mod that draws your project's agent loop live in a pane beside the chat whenever Claude runs it (a skill or command that plans, sends subagents, tests, reviews and loops back). The pane shows Now (the step at work, its time and tokens, each active agent with its model and context use, long-running shell commands, and waits on you), the path this run took with each step's time, tokens and outcome, the full map with arcs taken and red unexpected jumps, and alerts such as failing tests, bad verdicts, agents near full context, silent agents, repeat passes or a cold cache. It moves only on marker commands (node .claude/loop-graph/loop-step.mjs <step>) the loop runs at each step. The bundled /loop-graph:setup skill reads your loop, shows a draft, and on your go-ahead writes .claude/loop-graph/map.json, copies the marker script, and adds marker lines. /loop-graph opens the pane; reload and reset refresh it. Needs Claude Code 2.1.296+ and a one-time setup per project. Scores are documentation-calibrated, not tested by us; the preview is illustrative.
Who it is for
- People running plan-implement-test-review loops as skills or commands
- People who want to see live what each subagent is doing and where it stalls
Requirements
- Claude Code ≥ 2.1.296
- Run /loop-graph:setup once per project to create the map and markers.
- Each loop step runs a node marker command, so Node.js is needed.