Make a local capability callable
I created a local video-generation tool operated by an AI agent through a custom Model Context Protocol (MCP) server. It runs locally and was not exposed as an authenticated internet service.
03 / DEVELOPER SYSTEMSA collection of practical tools
Give an agent a useful tool. Give the workflow a clear purpose.
Developer tooling · Internal automation
TOOLS THAT DO THE WORK.
agent / workspace
// from instruction to action
agent → connect tools
video-generator / MCP
resolve-merge-conflicts / skill
code-analysis / hook
↳ ready for the next useful thing
MCP SERVERS
SKILLS & AUTOMATION
SKILLS & AUTOMATION
01 / THE CHALLENGE
Connect AI to work worth doing.
AI becomes more useful when it can participate in a real workflow. I built tools and instructions around tasks I actually encountered: generating video locally, resolving merge conflicts, analysing code and reviewing applications against a job description.
02 / THE DECISIONS01 02 03
Capture repeatable coding work
A PR merge-conflict resolution skill gives coding agents task-specific instructions. An automatic code-analysis hook helps identify low-quality code during agent-assisted development. I also use custom skills and hooks in daily coding and management work.
Put AI into a broader recruitment pipeline
I built a custom applicant tracking system from scratch using n8n and the OpenAI API. CV analysis against the applied-for job description was one of several pipeline steps and filters. The pipeline processed 400 applicants in two weeks.
03 / THE RESULT
What came out of it.
These projects connect AI to concrete, repeatable tasks. They include local developer tools, instructions used in daily work and a recruitment workflow that handled a real batch of applications.
The terminal visual is an illustration of the tooling collection, not a live terminal or an execution log.
WHAT I TAKE FORWARD“Useful automation starts with understanding the task.”