8000 truSwen (Benjamin Zsigri) · GitHub
[go: up one dir, main page]

Skip to content
View truSwen's full-sized avatar

Block or report truSwen

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
truSwen/README.md

Benjamin Zsigri

AI-Assisted Product Builder | AI Automation & Agent Workflow Specialist

I'm the founder of Tribloc. I turn business problems into working prototypes, automations and internal tools.

I define the requirements, system boundaries and acceptance criteria, direct AI coding agents such as Claude Code and Codex, and check the result through tests, review and documented limits. I have about one year of hands-on practice. I do not code independently: AI tools write most of the code, to my requirements and checks.

What I work on

  • Prototypes and internal tools: from a loosely defined business problem to a scoped, testable result with a clear handoff.
  • Automation and integration: workflows across APIs and MCP tools, using Make, n8n and code-based components built with AI agents.
  • Agent workflows: task scoping, isolated work areas, approval gates, review steps and failure handling.
  • Quality and cost control: acceptance tests, evaluation metrics, explicit unknown states and usage-cost tracking.

Public demos

All of these run on synthetic data. None is a client rollout.

  • Synthetic MCP Tool Demo (live): real MCP client and server with three read-only tools and 36 automated tests.
  • Evidence-Gated Workflow (live): approval gates, evidence checks, bounded retries and duplicate-action prevention. No live CRM or LLM connection.
  • AI Usage & Cost Explorer (live): exact cost estimates and a replay-aware local ledger, with 49 documented tests. No real spending.
  • AI Evaluation Lab (live): threshold trade-offs, error rates and explicit measurement rules, with 39 documented tests.
  • Tribloc Project Fit Assistant: a public Agentplace agent checked against nine scenarios. I set the goal and acceptance criteria; Claude operated the platform and ran the tests.

My own systems (private code)

These are my own Tribloc projects. The descriptions separate what is implemented from what is still in progress.

  • Agent Router: routing and workflow-control components, workspace isolation and review gates for parallel coding agents, used on my own repositories. This does not establish a fully deployed autonomous swarm.
  • Tribloc Lead Intelligence: Python company-research components with source-backed evidence. Independent quality validation remains in progress; no accuracy or conversion guarantee is claimed.
  • Tribloc OS: an operations backend with implemented development and staging paths. Full business cutover is not finished.
  • Write It Once: a public browser proof of concept exploring reusable form input and a passkey interaction. It is a prototype, not a production identity or government-service platform.

Read the public case studies on Contra.

Tools I work with

Claude Code · OpenAI Codex · Make · n8n · Hatchet · GitHub · MCP, plus AI-assisted work with Python, TypeScript, Cloudflare and Supabase.

Working approach

I start with a defined problem, access boundaries and acceptance criteria. Implementation, testing, review, deployment and live verification are separate steps, with human approval for consequential actions. Documented historical tests are not presented as a fresh independent audit or proof that every planned feature is live.

Work with me

Available for remote, paid project work: working prototypes, business applications, automations, integrations, AI workflows and internal tools, with agreed scope and deliverables.

Popular repositories Loading

  1. truSwen truSwen Public

    Config files for my GitHub profile.

  2. ergvein ergvein Public

    Forked from hexresearch/ergvein

    Multiplatform BTC wallet with client side filters, atomic swaps and coin mixing

    JavaScript

  3. carego carego Public

    carego files

    HTML

  4. missing-bits-solver missing-bits-solver Public

    TypeScript

  5. carego_csomagkovetes carego_csomagkovetes Public

    HTML

  6. FuzzyAI FuzzyAI Public

    Forked from cyberark/FuzzyAI

    A powerful tool for automated LLM fuzzing. It is designed to help developers and security researchers identify and mitigate potential jailbreaks in their LLM APIs.

    Jupyter Notebook

0