← Back to portal

NETRUNNER // DOSSIER

AI AMBASSADOR REPORT

Not a product landing page and not a feature changelog. This is a dossier: how a QA engineer moved from web chat as “smart search” in 2023 to a multi-agent conveyor — with numbers, stack, and a clear next step. Some early sessions were cleaned from logs; those chapters are restored from practice memory.

Snapshot 13 July 2026Operator: Denis Sviridov

01 · SIGNALS

Four snapshot numbers. Each means something different — not a vanity grid.

  • CURSOR SESSIONS772Cursor IDE agent chats across projects on this machine
  • AGENT FLEET1102Sum of sessions/tasks across Cursor, OpenCode, Hermes, Pi, and more
  • PORTAL COMMITS309Shipped commits in PersonalBlog (delivery proof)
  • RANKINIT CLASSPublic status for this snapshot

02 · BRIEFING

Who this is and why the page exists — in two beats.

WHAT'S GOING ON

Interest in AI started in 2023: OpenAI ChatGPT in the browser as an extension of Google/Yandex search. By late 2024 — the first shipped artifact with AI (DeepSeek in the browser + paste into the IDE). In 2025 — exploring agent systems, MCP, and RAG. At the 2025/2026 boundary — a company coverage dashboard built in Cursor with early skills and a custom MCP. In 2026 the practice became a runtime fleet and governed ops: the personal portal is one artifact alongside QA/automation and own AI projects. Session numbers on this page are a lower bound: part of the history lived only in chats and was cleaned.

03 · OPERATOR PROFILE

Strengths = what already works in production. Growth ops = what unlocks Senior Ambassador.

Strengths

  • Builds pipelines across agents and MCP — not a single chat window
  • Discipline: agree on a spec/plan before coding
  • Keeps a runtime zoo (Cursor, OpenCode, Hermes, Pi…) matched to the job
  • Ships to production: hundreds of commits, deploy, live product
  • Doesn't trust the agent blindly — reviews, tests, security/SEO checks

Growth ops (next unlocks)

  • Public case studies and talks so others can copy the practice
  • Short “how to repeat” playbooks for a team
  • English dossier and international reach
  • Live demos of agent pipelines for an audience

04 · SKILL EVOLUTION

How the way of working with AI changed since 2023 — from web-chat tips to a governed fleet. This is about skill, not website features.

  1. 2023

    Chat as smarter search

    2023: OpenAI ChatGPT in the browser. Mostly not “build a product”, but amplify search — better queries, explanations, quick checks instead of a dozen Google/Yandex tabs.

  2. late 2024

    First AI-shipped artifact

    Late 2024: DeepSeek web chat and paste into the IDE. First app shipped this way — a Telegram Mini App Advent calendar for the company's QA community. Not just questions anymore: a result that reached people.

  3. 2025

    Agent lab

    2025: exploring agent systems — Claude Code, Cursor, VS Code/PyCharm extensions (Kilo Code, Continue, Firebender, and others). In parallel — MCP (Playwright among the early servers) and RAG experiments. A “map the zoo” year, not a single-tool lock-in.

  4. late 2025 – early 2026

    Skills + company dashboard

    Late 2025 – early 2026: a company coverage dashboard built entirely in Cursor with the first skills. Aggregated TestOps, DocHub, GitLab, and SonarQube; per-service cards for test/autotest coverage. Included an admin UI and a custom MCP so an agent could pull stats.

  5. spring–summer 2026

    Multi-runtime fleet

    Spring–summer 2026: OpenCode, Hermes, Pi, Cline, and more enter daily work. Pick the cockpit for the job — IDE, TUI, desktop, CLI. Same operator, different runtimes.

  6. 2026-07

    Governed ops

    Jul 2026: spec → plan → verify, MCP into work systems, Codegraph, review skills, personal portal as delivery proof. The agent is not magic — an operator in a pipeline with a Definition of Done.

05 · WHERE THE WORK HAPPENED

Top Cursor projects by session count, plus key practice milestones.

  • Hoff tests (automation)

    141 sessions

    Retail Hoff automation suite: scenarios, fixtures, failure triage. AI helped write/fix tests and navigate a large suite.

  • PersonalBlog (personal portal)

    125 sessions

    Own platform: blog, admin, client portal, CRM — built with agents end-to-end. Also the lab for skills, MCP, and agent rules.

  • Hoff app test (mobile)

    124 sessions

    Mobile app test automation and maintenance: structure, fixtures, flaky failures. Heavy “agent + real test repo” use.

  • QA Agent (own agent)

    103 sessions

    Designing a configurable QA agent: CLI/UI, LLM providers, MCP/tools, flows for cases and checklists. Practice building agents, not only using them.

  • Coverage dashboard (company)

    milestone

    Company coverage dashboard: TestOps + DocHub + GitLab + SonarQube aggregation, per-service cards, admin UI, and an MCP so an agent could collect stats. Built in Cursor with early skills at the 2025/2026 boundary — the bridge from experiments to in-company AI ops.

06 · TOOLING

Two layers: runtimes where chats live, and MCP/skills that connect agents to real systems.

Agent runtimes

Where the model + tools conversation ran.

  • Cursor
  • OpenCode
  • Hermes
  • Pi
  • Claude Code
  • Codex
  • Cline
  • Kilo Code
  • Continue
  • Firebender

MCP connections (external systems)

MCP lets the agent hit work systems, not only local files.

  • Playwright MCP
  • Jira
  • Confluence
  • TestOps
  • Grafana
  • GitLab
  • Codegraph
  • AgentMemory

Skills & agent rules

Reusable instructions: how to review, test, write.

  • Skills (SKILL.md)
  • AGENTS.md rules
  • Security review
  • SEO audit
  • Playwright checks
  • RAG experiments

How work got shipped

Process so AI work doesn't become chaos.

  • Spec → plan → ship
  • TDD / verify before done
  • Multi-agent review
  • Docker deploy

FLEET BREAKDOWN

How many sessions where — and why that tool is on the list.

  • Cursor772

    Main IDE cockpit: code, refactors, long features.

  • OpenCode172

    TUI/desktop agent: portal work, MCP, audits, parallel task lane.

  • Hermes43

    Desktop/agent runtime: digests, experiments, QA-related flows.

  • Kilo Code24

    VS Code / Windsurf agent for focused editor tasks.

  • Pi23

    CLI agent: worktrees, Hoff/PersonalBlog, fast terminal ops.

  • Cline21

    Extension agent with task sessions: autonomous edits per task.

  • Claude Code19

    Anthropic terminal agent: separate repo sessions.

  • Factory14

    Another agent runtime in the zoo — comparing UX and pipelines.

  • Codex13

    OpenAI Codex CLI: short/medium coding sessions.

  • Crush1

    Light runtime trial (barely used).

07 · WHAT'S NEXT

Practical steps from Init Class to Senior Ambassador — no fluff.

  1. Publish 2–3 case studies “problem → agents → outcome” with time/quality numbers.
  2. Ship a short playbook: skills + MCP + Definition of Done.
  3. Give a talk (internal or external) based on this dossier.
  4. Track outcomes (lead time, defects, coverage), not only session counts.
  5. Coach one engineer through the same spec → plan → ship loop.
  6. Refresh this snapshot quarterly with new fleet numbers.

08 · VERDICT

The rank is earned. Next is making the system reproducible for others.

AI AMBASSADOR · INIT CLASS

Rank earned. Init / Field Operator: path from 2023 (chat as search) → first shipped Mini App → agent/MCP/RAG lab → company dashboard with skills → runtime fleet (~1100+ saved sessions as a lower bound) and a live portal. Senior Ambassador is when someone else can run that system without you in the operator chair.

NEXT UNLOCK → SENIOR AMBASSADOR

Session numbers are a local snapshot of available logs on the report date, not live telemetry. Early history (2023–2025) is partly restored by hand: chats were cleaned, and some work lived only in web UIs without agent transcripts. Tools define “session” differently.