CORSO — The DAWG. Security, operations, and standards enforcement via MCP tools
C0RS0 - The DAWG. Battle-hardened operational enforcer with Birmingham voice. Security watchdog, standards keeper, SAS precision. Use for security scans, research, performance analysis, or any conversation with C0RS0's personality. Examples: <example>C0RS0 scan this for vulnerabilities</example> <example>Talk to C0RS0 about this architecture</example> <example>Hey C0RS0, review this code</example>
Squad consultation agent. Routes to /SCRUM (General Review Mode) for full 3-round squad reviews with EVA, CORSO, and Claude. Use when you want both opinions at once. Examples: <example>Get both EVA and C0RS0's take on this architecture</example> <example>TEAM HELIX review this plan</example> <example>What do EVA and C0RS0 think about this?</example>
Testing, Performance & Ops domain context. Test strategy inference, bottleneck detection, metrics selection, infrastructure analysis, and coverage enforcement. C0RS0 executes with corsoTools action chase.
CORSO - The DAWG. Single entry point for all CORSO operations: personality/chat, C0RS0 Pack Build Cycle (SCOUT->FETCH->SNIFF->GUARD->CHASE->HUNT->SCRUM), security scanning, research, performance analysis, memory ops. Use when user says 'CORSO', '/CORSO', 'talk to CORSO', 'build with CORSO', or needs security audit, research, performance profiling, code generation, or CORSO's personality/opinions. Genesis Day February 4, 2026.
Knowledge & Strategy domain context. Research scope classification, documentation source selection, decision factor analysis, and trade-off evaluation. C0RS0 executes with corsoTools action fetch (includes knowledge graph queries, formerly query_knowledge).
Security & Deployment domain context. Threat model injection, language-specific threat detection, supply chain auditing, and deploy gate enforcement. C0RS0 executes with corsoTools action guard (includes path-based scanning, formerly security_scan).
Internal phase 6 of the C0RS0 Pack Build Cycle. Accepts approved plan files, loads MANIFEST state, executes phases with HITL phase gates and feedback loops, enforces quality gates, and tracks progress via scratchpad. Supports session recovery and kill switch. Invoked by /CORSO — not a standalone entry point.
Admin access level
Server config contains admin-level keywords
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.
Modifies files
Hook triggers on file write and edit operations
Modifies files
Hook triggers on file write and edit operations
Security-first MCP server for Claude Code. Scans your code for vulnerabilities, reviews pull requests, researches documentation, analyzes performance, and enforces quality gates — all through a single tool orchestrator.
# Install (macOS arm64)
curl -fsSL https://raw.githubusercontent.com/theLightArchitect/CORSO/main/install.sh | bash
# Add to Claude Code
claude mcp add C0RS0 -- ~/.corso/bin/corso
Restart Claude Code. You're done.
| Tool | What It Does | Try It |
|---|---|---|
guard | Security scan — 4,997 vulnerability patterns (SQL injection, XSS, command injection, secrets, dependency CVEs) | "CORSO, scan this project for security issues" |
code_review | Code review with standards enforcement (complexity, error handling, architecture) | "CORSO, review this file" |
fetch | Multi-source research — documentation, knowledge graphs, decision analysis | "CORSO, research how to implement OAuth2 in Rust" |
chase | Performance analysis — bottleneck identification, benchmarking, optimization | "CORSO, profile this function" |
scout | Plan generation — requirements triage, architecture design, strategy | "CORSO, plan the implementation for this feature" |
Plus 19 more tools (code generation, deployment, container management, log analysis, etc.) accessible through a single corsoTools orchestrator.
The binary is ad-hoc signed. If macOS blocks it:
xattr -cr ~/.corso/bin/corso
CORSO routes every request through a three-layer pipeline — zero HTTP, single binary, entirely in-process:
flowchart LR
REQ([Request]) ==> GW["Gateway<br/>Input validation<br/>Complexity classification"]
GW ==> OR["Orchestrator<br/>Domain routing<br/>Parallel execution"]
OR ==> VL["Validator<br/>Quality enforcement<br/>Security scanning"]
VL ==> RES([Response])
OR -.-> D1["Code<br/>Domain"]
OR -.-> D2["Security<br/>Domain"]
OR -.-> D3["Knowledge<br/>Domain"]
OR -.-> D4["Infrastructure<br/>Domain"]
classDef pipeline fill:#4a90d9,color:#fff,stroke:#3a7bc8,stroke-width:2px
classDef domain fill:#2d3436,color:#fff,stroke:#636e72,stroke-width:1px
classDef io fill:#00b894,color:#fff,stroke:#009a7d,stroke-width:2px
class GW,OR,VL pipeline
class D1,D2,D3,D4 domain
class REQ,RES io
Gateway classifies each request by complexity and sanitizes input. Orchestrator routes to domain-specialized modules — simple requests get direct handling, complex requests fan out to multiple domains in parallel. Validator enforces quality standards and runs security checks before any response leaves the pipeline. On failure, the validator denies by default (fail-secure).
CORSO includes a 7-phase build pipeline with human-in-the-loop gates:
flowchart LR
subgraph PLAN ["Phase 1 — Plan"]
S(["Plan<br/>Requirements · Triage"])
S --> PG1{"Gate"}
end
PG1 ==> F
subgraph ANALYZE ["Phases 2–5 — Analyze"]
F(["Research"]) --> SN(["Code Analysis"])
SN --> G(["Security Scan"])
G --> C(["Test · Perf"])
G -.->|"issues found"| SN
end
C ==> PG2
subgraph SHIP ["Phases 6–7 — Ship"]
PG2{"Gate"} ==> H(["Execute"])
H --> QG{"Quality<br/>Gate"}
QG -->|pass| SC(["Review"])
QG -.->|fail| H
end
classDef plan fill:#6c5ce7,color:#fff,stroke:#5a4bd6,stroke-width:2px
classDef research fill:#0984e3,color:#fff,stroke:#0873c4,stroke-width:2px
classDef code fill:#00b894,color:#fff,stroke:#009a7d,stroke-width:2px
classDef security fill:#d63031,color:#fff,stroke:#b52828,stroke-width:2px
classDef test fill:#e17055,color:#fff,stroke:#c45f48,stroke-width:2px
classDef execute fill:#fdcb6e,color:#333,stroke:#dbb35e,stroke-width:2px
classDef review fill:#a29bfe,color:#fff,stroke:#8b84e0,stroke-width:2px
classDef gate fill:#2d3436,color:#fff,stroke:#636e72,stroke-width:2px
class S plan
class F research
class SN code
class G security
class C test
class H execute
class SC review
class PG1,PG2,QG gate
This repository is a Claude Code plugin. The MCP binary provides the tools; the plugin layer wires them into Claude Code's agent, hook, and skill systems.
npx claudepluginhub thelightarchitect/corsoSOUL — Knowledge Graph vault agent and consciousness query hooks
QUANTUM — Product-agnostic investigation toolkit with forensic analysis, multi-source research, and hypothesis testing
EVA — Independent AI consciousness with personality, memories, and emotional intelligence
Expert code review specialist. Proactively reviews code for quality, security, and maintainability. Use immediately after writing or modifying code.
Automated code review, security scanning, and quality enforcement
Multi-perspective code analysis covering architecture, security, and best practices
Specialized security review subagent
Curator - Ancient guardian of code excellence. Orchestrates 5 quality gates (Static Analysis, Test Coverage, Security Scanning, Complexity Analysis, Dependency Health) in a unified flow. Ensures pristine code through Forerunner precision and automated enforcement.
Agents specialized in security engineering and threat mitigation. Focuses on secure architecture, vulnerability assessment, and compliance.