By felimet
Unified knowledge graph layer combining GitNexus (code architecture) and graphify (documents/knowledge). Auto-detects inefficient searches and suggests /cortex; PreToolUse hook injects graph context before every Grep/Glob/Bash op.
Traces what breaks or changes when a symbol, module, or API is modified. Uses GitNexus AST-precise dependency graph.
Indexes current (or specified) repo into both GitNexus and graphify simultaneously.
Routes a natural-language question to the right graph tool and returns a synthesized answer.
Routes codebase and domain questions to the right tool: **GitNexus MCP** for code architecture, **graphify CLI** for documents and domain knowledge.
Claude Code plugin unifying GitNexus (code architecture) and graphify (documents/knowledge) into a single /cortex interface.
/cortex when graph traversal beats recursive grep/find# GitNexus
npm install -g gitnexus
claude mcp add gitnexus -- npx -y gitnexus@latest mcp
# graphify
uv tool install graphifyy
graphify claude install
Add to ~/.claude/settings.json:
{
"extraKnownMarketplaces": {
"repo-cortex": {
"source": {
"source": "github",
"repo": "felimet/claude-repo-cortex"
}
}
}
}
Then: claude plugin install repo-cortex
claude plugin install /path/to/claude-repo-cortex
After installing the plugin, add the following to your ~/.claude/CLAUDE.md (global) or .claude/CLAUDE.md (project-level):
# cortex
- **cortex** (repo-cortex plugin) - unified knowledge graph: routes code-architecture
questions to GitNexus MCP tools, document/domain questions to graphify CLI. Trigger: `/cortex`
When the user types `/cortex`, invoke the Skill tool with `skill: "cortex"` before doing anything else.
## cortex proactive suggestions (always active)
Detect the following situations and proactively offer a one-line `/cortex` suggestion BEFORE
proceeding with the less efficient alternative.
Format: `[cortex] <situation> → <suggested command>`.
Do NOT suggest if `/cortex init` output files are absent AND repo is not in `gitnexus list`.
| Situation | Suggest |
|-----------|---------|
| "how does X work", "explain architecture", "what does this codebase do" | `/cortex query "X"` or read `graphify-out/GRAPH_REPORT.md` |
| "what calls X", "who uses Y", "where is Z referenced" | `/cortex impact X` |
| "what breaks if I change X", "impact of modifying Y" | `/cortex impact X` |
| "safe to rename X" | GitNexus `rename` + `shape_check` via `/cortex` |
| About to run multiple Glob/Grep to understand code structure | `/cortex query "<question>"` |
| User pastes a paper/doc URL | `/cortex add <url>` |
| Session starts in repo with no `graphify-out/` and not in gitnexus index | `/cortex init` (one-time) |
| End-to-end flow question spanning multiple modules | `/cortex query` + GitNexus `route_map` |
| Debugging a bug across multiple files | `/cortex impact <entry_symbol>` first |
Suggestion rules:
- One line only. No multi-paragraph explanation.
- Offer suggestion, then wait. Do not auto-execute `/cortex` commands.
- If user declines or ignores, proceed — do not repeat for same question.
- If both graphs absent and repo small (<10 files), skip.
| Skill | Trigger | Purpose |
|---|---|---|
cortex | /cortex | Unified entry point + routing |
cortex-init | /cortex-init | Build both graphs for current repo |
cortex-impact | /cortex-impact <symbol> | Impact analysis via GitNexus |
cortex-query | /cortex-query <question> | Smart-routed graph query |
| Hook | Trigger | Purpose |
|---|---|---|
| SessionStart | Every session | Inject proactive suggestion rules |
| PreToolUse | Grep/Glob/Bash | Inject graph context when available |
| PostToolUse | Bash | Suggest /cortex for recursive grep/find/cat patterns |
cd your-project
/cortex-init # build both graphs (one-time)
/cortex-query "how does the auth flow work"
/cortex-impact authMiddleware
/cortex add https://paper-url # add external knowledge
MIT
This plugin is built on top of two excellent open-source projects:
Executes bash commands
Hook triggers when Bash tool is used
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
claude-cogsec — Cognitive Security hardening pack for Claude Code. Derived from the cogsec / affaan-m guides (https://github.com/affaan-m/ECC#the-guides). Covers token optimization, memory persistence, evaluation, parallelization, and agentic security (CVEs, sandboxing, sanitization, AgentShield). Includes PostToolUse observability hook, deny-list templates, and the /cogsec-audit slash command.
Structured workspace management for MATLAB MCP projects. Enforces directory conventions, session protocols, cross-session continuity via README state machines, and integrates with the matlab-skills plugin (live-script, test-creator, test-execution, performance-optimizer, uihtml-app-builder, digital-filter-design). This skill handles WHERE files go; matlab-skills handles HOW code is written.
npx claudepluginhub felimet/claude-repo-cortex --plugin repo-cortexPermanent coding companion for Claude Code — survives any update. MCP-based terminal pet with ASCII art, stats, reactions, and personality.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Intelligent prompt optimization: injects the right context at the right moment so Claude lands a better first output. Clarifies vague prompts with research-based questions, plus targeted nudges for approach selection, plan readability, workflow routing, background execution, subagent routing, output readability, user-decision questions, and plan-mode assessment
Memory compression system for Claude Code - persist context across sessions
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Claude harness - A harness for solo developers (Vibecoders) to handle full-cycle contract development.