How epvh-commits/t3mp3st works
T3MP3ST is a self-hosted, multi-agent offensive-security framework. An operator submits an authorized target through the CLI or War Room; Op General plans the mission, operators perform tool-backed testing, and evidence gates and reporting support findings and disclosure. The README also documents CTF, source-code, robotics/OT, smart-contract, cloud, mobile, and binary capabilities. Some of those implementations are not sampled, so their wiring is shown only where the tree or README supports it.
Main entry points: operators/index.ts Operator agents · server.ts War Room API · general/index.ts Op General · llm/index.ts LLM backbone
Components 20
User entry
- CLI
src/cli.ts - War Room API
src/server.ts
Mission control
- Op General
src/general/index.ts - Mission engine
src/mission/index.ts - Orchestrator
src/orchestration/orchestrator.ts - Operator agents
src/operators/index.ts
Testing capabilities
- Recon and attack graph
src/recon/attack-graph.ts - Tool arsenal
src/arsenal/index.ts - CTF executor
ctf/executor/ctf_executor.py - Python code ingest
src/recon/code-ingest.ts - Other domain capabilities
src
Evidence and disclosure
- Evidence gate
src/evidence/gate.ts - Bounty integration
src/integrations/bounty.ts
Agent support
- LLM backbone
src/llm/index.ts - Operator prompts
src/prompts/index.ts - Security resources
src/resources/index.ts
Other components
- Security operator
- Configured model or local agent
- Authorized target
- Bounty platform
Connections 19
- Security operator → CLI: submits mission
- Security operator → War Room API: submits mission
- CLI → Op General: requests plan
- War Room API → Op General: accepts directives
- War Room API → LLM backbone: initializes
- Op General → LLM backbone: plans with
- Op General → Mission engine: executes plan
- Mission engine → Orchestrator: coordinates
- Orchestrator → Operator agents: dispatches work
- Operator agents → LLM backbone: reasons with
- Operator agents → Operator prompts: uses prompts
- Operator agents → Tool arsenal: invokes tools
- Tool arsenal → Authorized target: tests
- Operator agents → Recon and attack graph: performs reconnaissance
- Operator agents → Evidence gate: gates findings
- War Room API → Bounty integration: submits findings
- Bounty integration → Bounty platform: submits report
- LLM backbone → Configured model or local agent: sends requests
- Mission engine → Security resources: offers resources