hermes agent challenge · 2026

built with hermes agent + ollama + local llm

Escape
Tutorial Hell
for Good

Hermes Mentor scans your GitHub, understands your real skill gaps, and builds you a personalized project roadmap — powered by a fully local LLM. No more watching. Start building.

see it in action → view on github
0
repos analyzed
0
skill gaps found
0%
100% local · private
0
skills learned
hermes-mentor — wsl2 terminal
hermes run "analyze my github repos and create a learning roadmap"
✓ Connecting to GitHub API...
Found 14 repositories · Analyzing code patterns...
⚡ Skill gap detected: No production API work found
⚡ Skill gap detected: Missing CI/CD pipelines
⚡ Skill gap detected: No testing frameworks used
✓ Roadmap generated · 4 weeks · 6 projects
📬 Sent to Telegram · First challenge delivered
Skill saved: github-audit-v1.md

From tutorial watcher
to real builder

Hermes Mentor uses every agentic superpower Hermes ships with — memory, skill learning, cron jobs, and Telegram messaging — to be the mentor you never had.

01 / SCAN
🔍

GitHub Audit

Hermes hits the GitHub API, reads every repo you've built, and builds a real picture of your actual skill level — not what you claim.

02 / THINK
🧠

Gap Analysis

Local Llama 3 via Ollama reasons across your code patterns, languages used, and project complexity to identify exactly what's missing.

03 / PLAN
🗺️

Roadmap Generation

Not generic tutorials. Hermes generates a 4-week project ladder tailored to your gaps — each project closes a specific weakness.

04 / PUSH
📬

Daily Telegram Nudges

Every morning, a Hermes cron job fires off your daily challenge, a hint if you're stuck, or a "you did it" when you push code.

05 / LEARN
📈

It Gets Smarter

After every session, Hermes saves a skill file about you. Next week's roadmap knows what you struggled with last week.

06 / OWN
🔒

Fully Local & Private

Ollama runs the LLM on your machine. Your GitHub data, your learning history, your skill files — nothing leaves your WSL2 box.

Built on every Hermes
superpower

PERSISTENT MEMORY

It remembers you, always

Hermes writes a USER.md and MEMORY.md about your coding habits, preferred languages, past struggles, and wins. Every session starts informed, not blank.

~/.hermes/memory/ USER.md MEMORY.md FTS5 search
SKILL LEARNING

Gets smarter every run

After each GitHub audit, Hermes converts what it learned about you into a reusable skill file. Future roadmaps improve automatically.

agentskills.io GEPA loop skill curator
CRON SCHEDULER

Daily accountability, automated

A scheduled cron task fires every morning — checking if you pushed code, generating the next challenge, and sending your daily nudge to Telegram.

hermes cron 08:30 daily progress tracking
LOCAL LLM · OLLAMA

Zero cloud. Zero cost.
100% yours.

The LLM that powers Hermes Mentor runs entirely on your machine through Ollama. No API keys, no monthly bills, no data sent to a third party. Your learning journey stays private.

🦙 ollama · localhost:11434/v1 · qwen2.5-coder:7b
Ollama qwen2.5-coder llama3.2 mistral WSL2 Python 3.11
🐙

GitHub API

Reads your repos, commits, languages

🤖

Hermes Agent

Orchestrates tools, memory, skills

🦙

Ollama LLM

Local inference, no cloud

data flow
📱

Telegram

Daily nudges and challenges

🧠

Memory Store

Your skill profile, sessions

📚

Skill Files

Reusable patterns, learnt

See Hermes Mentor
in action

// Generated roadmap for TheCoderAdi · Based on 14 repos analyzed

W1

REST API from scratch — no tutorials

Build a full CRUD API with authentication. Your repos show strong React skill but zero backend API code. This project closes that gap.

✓ completed Node.js / FastAPI closes: API gap
W2

Add CI/CD with GitHub Actions

None of your 14 repos have any pipeline config. Build a project and wire it to GitHub Actions — test, lint, deploy on every push.

← in progress GitHub Actions closes: CI/CD gap
W3

Write tests for an existing project

Go back to a past project and add unit + integration tests. Hermes scanned zero test files across all your repos. This fixes that.

upcoming Pytest / Vitest closes: testing gap
W4

Deploy a real thing to production

Ship something live. End-to-end: code, tests, CI, deploy. Hermes will review your GitHub activity and verify it's actually online.

upcoming Docker / Railway closes: deploy gap

// Hermes GitHub audit · github.com/TheCoderAdi

TheCoderAdi/GoPay-UI
React · strong
TheCoderAdi/WeatherApp
RN · solid
TheCoderAdi/GamingBeast
no tests
14 repos · CI/CD pipelines
0 found
14 repos · backend API code
minimal
14 repos · test files
0 found
→ Hermes verdict: Strong frontend skills. Critical gaps in testing, CI/CD, and backend API development. 4-week roadmap generated.

// Daily message sent via Hermes cron · 08:30 every weekday

HM
Hermes Mentor
online
Good morning Aditya! ☀️

Week 2 · Day 3 check-in

I noticed you haven't pushed to ci-cd-project since Tuesday. Here's a hint to get unstuck:

Your GitHub Actions workflow is failing because the node version isn't pinned. Add node-version: '20' to your setup-node action.

You've got this 💪
08:30
thanks! fixing it now
08:47
🎉 Build passing!

Just saw the green check on your repo. Nice work. You've officially closed the CI/CD gap.

Week 3 challenge unlocked: write tests for GoPay-UI. I'll send details tomorrow morning.
09:12