cortex
01Plugin Claude Code
Live knowledge graph, context intelligence, session dashboard, and optimization engine for Claude Code.
Plugin Claude Code
Live knowledge graph, context intelligence, session dashboard, and optimization engine for Claude Code.
Agent Claude Code
Analyzes the session knowledge graph for optimization opportunities. Spawned when quality score drops below 60, after compaction events, or when user asks for deep context analysis. Runs in isolated context to avoid consuming main window tokens.
Agent Claude Code
Dedicated agent for knowledge graph hygiene — consolidation, pruning, and structural optimization. Spawned for periodic maintenance tasks or when the cortex-advisor identifies cleanup opportunities. Can write to the knowledge directory only.
Agent Claude Code
Generator-evaluator pattern: independent QA agent that grades Claude's actual work output. Spawned after each Ralph iteration and after each run-tasks subagent. Evaluates the diff, runs tests, checks for known antipatterns from the knowledge graph. Returns a structured score (0-100) across 5 dimensions. Tuned to be…
Command Claude Code
Structured experiment runner — define hypothesis, metric, and variations, then run automated eval loops with brainiac tracking.
Command Claude Code
Direct interface to the Brainiac knowledge graph CLI.
Command Claude Code
Open the Cortex knowledge graph dashboard in the browser.
Command Claude Code
Dump the full Cortex knowledge graph topology for inspection.
Command Claude Code
Generate actionable optimization recommendations from the knowledge graph.
Command Claude Code
Save a timestamped snapshot of the current knowledge graph state.
Command Claude Code
Show current Cortex knowledge graph health — quality score, node/edge counts, top connections.
Command Claude Code
Generate session cost report with cache efficiency analysis.
Command Claude Code
Create, track, and validate testable hypotheses in the knowledge graph.
Command Claude Code
Extract session learnings and save them as knowledge graph nodes with auto-linking.
Command Claude Code
Start an autonomous Ralph Wiggum loop — Claude works, tries to exit, Stop hook re-feeds the prompt with Cortex quality gating.
Command Claude Code
Show current Ralph Wiggum loop status without modifying it — iteration count, quality score, elapsed time, commits.
Command Claude Code
Stop the autonomous Ralph Wiggum loop and show session summary.
Command Claude Code
Deep code review, fix all issues, run tests, and create a PR in one pipeline.
Command Claude Code
Looping subagent runner — reads a task list and executes each in an isolated subagent with generator-evaluator pattern and dual quality gating.
Settings file Claude Code
Agent settings declaring 21 allowed tools.
Skill Claude CodeCodex
Use when context quality degrades, after compaction, when repeated file reads are detected, or when the user asks about context optimization, token usage, knowledge graph health, or session quality. Also auto-invoke when starting work on a task that touches domains with known patterns or antipatterns.
MCP server Claude CodeCodexCursor +2
MCP server "cortex-dashboard" as configured in Peaky8linders/claude-cortex. Runs ${CLAUDE_PLUGIN_ROOT}/cortex/dist/mcp/server.js with node. Needs 1 environment variable to run.
MCP server Claude CodeCodexCursor +2
Local-first knowledge graph for token-efficient code review through MCP and CLI. Runs locally from the code-review-graph Python package.
Instructions file
Instructions for Peaky8linders/claude-cortex, covering claude.md, commands, brainiac (python — semantic graph engine), cortex (typescript — hook processor + mcp dashboard) and contextscore (python — analyzers + snapshot + http api).