[ CAPABILITY CATALOG ]
Capability Catalog
Open-source components, skills, MCP servers, knowledge bases, and owned products serve as selection assets; delivery is organized by agent efficiency, cloud infrastructure, and AI transformation.
Agent Orchestration · 23AI Coding · 29Quality & Eval · 4Red Team · 8Knowledge Engine · 15Creative · 16Data Insight · 14Productivity · 14
[ BUSINESS-LINE DELIVERY ]
Business-line Delivery Packages
The open-source catalog is an asset base. Delivery is organized by three high-complexity lines: agent efficiency, cloud infrastructure, and AI transformation. We do not sell low-end hours; scope is evaluated by complexity, phase, and executable assets.
Agent Efficiency Architecture
Scoped by phase / no low-end outsourcing
For teams with agent, enterprise assistant, knowledge assistant, or multi-agent prototypes hitting memory, collaboration, performance, and operability limits.
- Four-layer memory, multi-agent collaboration, and agent harness architecture
- LLM gateway, queues, caches, microservice boundaries, and cost governance
- Architecture plan, runtime framework, evaluation / observability metrics, and delivery path
Cloud Infrastructure and Private Cloud
Scoped by asset size / phased delivery
For teams needing cloud exit, private cloud, local GPU / servers, hybrid cloud, security, monitoring, backups, and long-term operations.
- Public-cloud asset audit, cloud-exit path, cutover, and rollback planning
- Private cloud, server / GPU / storage, network security, and access-control design
- Topology, configuration list, monitoring / backup / DR plan, and operations boundaries
Intelligent Transformation Strategy
Scoped by organization / roadmap + pilots
For teams needing AI transformation direction, business-process redesign, organizational skill capture, and AI connected to operating metrics.
- Operating goal, workflow, data asset, knowledge, and scenario-priority diagnosis
- AI-native workflow, skill library, agent workspace, and operating dashboard design
- 30 / 60 / 90 / 180 day roadmap, pilot portfolio, and review metrics
Agent efficiency, cloud infrastructure, and transformation cover the core paths of AI system delivery.
Each line is broken into scenarios, cases, capabilities, delivery path, and executable outputs.
Open-source components, skills, MCP servers, knowledge bases, and owned products support selection and delivery.
OpenClaw Deployment 2.1k
12+ messaging platform unified AI assistant gateway. Supports WhatsApp, Telegram, Discord, iMessage and more. Local-first architecture, fully privacy-controlled.
Local-firstMulti-channelPrivacy-controlledProfessional Service
Langfuse Deployment 21.6k
Open-source LLM observability & evaluation platform: Trace, Prompt management, datasets & comparative evaluation. We provide private deployment & ops.
Private DeployTraceEval LoopProfessional Service
Promptfoo Regression Testing 10.3k
LLM Prompt/Agent regression testing framework, CI gate-ready. We provide test suites, metrics & pipeline integration.
CI GateRegressionTest SuitesProfessional Service
DeepEval Evaluation 13.5k
LLM evaluation & unit testing framework (metrics + LLM-as-judge). We provide metric systems, datasets & CI integration.
MetricsLLM JudgeCI IntegrationProfessional Service
Ragas RAG Evaluation 12.5k
RAG quality evaluation framework (faithfulness/relevance/answer quality). We provide dataset construction, metric interpretation & continuous evaluation.
RAG EvalDatasetsMetricsProfessional Service
Garak Red Team Scan 6.9k
LLM security red team scanner covering jailbreak, prompt injection, sensitive info leakage risks. We provide baseline scanning & remediation retesting.
Red TeamJailbreak/InjectionBaseline ReportRemediation
PyRIT Adversarial Testing 3.4k
Microsoft open-source adversarial testing orchestration framework for systematically generating attack cases & assessing defenses. We provide attack library customization & reports.
AdversarialAttack LibraryAssessment ReportProfessional Service
Maigret OSINT Lead Aggregation 27.4k
A mature open-source username OSINT aggregator: provide a username, check 3000+ sites for possible accounts, and organize public profile signals into reports. Built for authorized investigations and lead triage, not privacy abuse.
Username OSINT3000+ SitesRecursive LeadsReport Export
LGBT-Prompt Safety Corpus Analysis 452
Not a prompt-engineering project, but an adversarial prompt sample repo that treats identity-sympathy bias as an attack surface. Its value is safety research and defensive evaluation, not product building.
Jailbreak SamplesSafety EvalBias DefenseRed-team Research
Eigent Deployment 1.8k
Open-source multi-agent collaboration desktop app powered by CAMEL-AI. 35+ built-in toolkits with MCP integration. Build your own AI Workforce, locally deployed and privacy-controlled.
Multi-agentLocal DeployMCP IntegrationProfessional Service
RAGFlow Deployment 25.3k
Open-source RAG engine with deep document understanding. Supports complex format parsing, visual chunking, multi-path recall & traceable citations. Out-of-the-box knowledge base Q&A.
Deep Document ParsingVisual ChunkingMulti-path RecallProfessional Service
AnythingLLM Deployment 30.7k
Full-stack RAG application: decouples LLM / Vector DB / Embedder into pluggable modules. Workspace-level knowledge isolation, multimodal data pipeline, Desktop + Docker dual-form. Private deployment — data never leaves your domain.
Modular DecouplingWorkspace IsolationMultimodal PipelineProfessional Service
SciBot Scientific RAG Pattern 44
Not a modern agent system, but an early scientific RAG chatbot pattern: embed scientific papers and figures for retrieval, then pass relevant context into an LLM for domain-grounded answers.
Scientific RAGDocument RetrievalImage EmbeddingsMethod Pattern
Hello-Agents Agent Curriculum 50.9k
A systematic AI Agent tutorial from Datawhale, titled Building Agents from Scratch. It is not a production framework, but a structured curriculum for learning agent principles, code, tools, protocols, evaluation, and full cases.
Chinese CourseAgent TheoryHands-on CasesTraining
DocuSeal E-signature 16.7k
An open-source DocuSign alternative for self-hosted electronic signatures, giving teams control over PDF forms, signing workflows, file storage, email delivery, and API/webhook integrations. Requires license and legal-validity review before rollout.
E-signatureSelf-hostedPDF FormsAPI/Webhooks
GitNexus Code Graph 38.9k
A code knowledge graph engine for AI coding agents, structuring dependencies, call chains, module clusters, execution flows, and change-impact relationships for Claude Code, Codex, Cursor, and Windsurf.
Code GraphMCP ToolsImpact AnalysisMulti-repo Context
Understand-Anything Code Map 15.1k
A code-map generator for AI coding tools that turns codebases, docs, or knowledge bases into interactive knowledge graphs so humans and agents can quickly understand files, functions, classes, dependencies, and business flows.
Project MapOnboardingSemantic SearchDiff Impact
Graphify Project Memory Layer 49.2k
A project memory layer for AI coding assistants that turns code, docs, PDFs, images, videos, and SQL schemas into a queryable knowledge graph so Claude Code, Codex, Cursor, Gemini CLI, Hermes, and Kimi Code do not reread the whole repo every time.
Project GraphMultimodal InputMCP QueryTeam Shared
RTK Output Compression Proxy 50.1k
A CLI output compression proxy for AI coding tools. It filters, groups, truncates, and deduplicates outputs from git diff, rg, pytest, docker logs, kubectl logs, and more before Claude Code, Codex, Cursor, Gemini CLI, or Hermes sees them.
Token SavingsCommand HooksLog DedupRust Binary
Warp Agentic Terminal 59.0k
A modern terminal evolving into an agentic development environment, combining terminal workflows, code editing, Warp Agent, bring-your-own CLI agents such as Claude Code, Codex, Gemini CLI, OpenCode, and Oz-powered open-source workflows.
Modern TerminalMulti-agentCode EditingOz Workflows
OpenCLI CLI Standardization 199
The OpenAPI for CLIs — machine-readable descriptions of commands, parameters, options and exit codes. Productize internal CLI tools: one spec drives doc generation, auto-completion, and OpenCLI→MCP bridging. Pushed by the Spectre Console community, still in draft stage.
CLI ContractsMCP BridgeAuto DocsChange Governance
Strix Penetration Testing 20.6k
Fully automated open-source AI pentest tool — like a 24/7 red team hacker. Combines white-box source analysis with black-box dynamic exploitation across four agent phases. Doesn't just scan for vulns, it actually proves them.
Auto PentestWhite+Black BoxExploit VerificationMulti Agent
Dexter Financial Research Agent 16.4k
Autonomous agent for deep financial research. Transforms complex financial questions into structured research plans, auto-fetches data, analyzes & validates, then outputs trustworthy conclusions. Think first, act second, self-verify for reliability.
Autonomous ResearchMulti-source DataSelf-validationFinancial Analysis
Monty Secure Sandbox 5.7k
A minimal, secure Python interpreter written in Rust by the Pydantic team. Designed for AI embedded code execution. Microsecond startup, sandboxed isolation — lighter and faster than full containers.
Rust-builtμs StartupSecure SandboxEmbeddable
Not sure which line fits?
Share your agent prototype, cloud bill, current architecture, or workflow. We will first identify the right business line.