cd /news/developer-tools/introducing-devpub-open-source-dev-t… · home topics developer-tools article
[ARTICLE · art-82930] src=dev.to ↗ pub= topic=developer-tools verified=true sentiment=↑ positive

Introducing DevPub - Open Source Dev.to CLI Tool

A developer built DevPub, an open-source CLI tool for Dev.to that leverages over 40 API endpoints, including semantic search, analytics, and trend discovery, which existing tools ignore. The tool, built in a day, uses Gemini embeddings for semantic search and includes features like article validation and rate limiting. The developer also discovered undocumented API endpoints, such as the V1 Accept header and nested analytics responses.

read6 min views1 publishedAug 1, 2026

Recently I went looking for a CLI tool to manage my Dev.to articles from the terminal. I write 4-5 articles per month, track analytics obsessively, and wanted a git-backed workflow.

I found 9 existing tools. Tried them all. Here's what happened:

Every single tool does the same thing: publish an article. That's it. Maybe pull. Maybe validate tags.

Meanwhile the Dev.to API has 40+ endpoints including analytics, semantic search, ML-powered content concepts, follower engagement, trend tracking, and reading list management. Nobody uses them.

So I built ** devpub**.

devpub push -f articles/my-post.md
devpub pull

devpub stats

devpub dashboard

devpub search "building serverless apps" --semantic

devpub trends

devpub validate

The difference isn't one feature. It's coverage. Here's the comparison:

Capability devpub Everyone else
Publish/update articles Yes Yes
Pull articles to local Yes Some
Analytics (7 endpoints) Yes No
Semantic search Yes No
Trend discovery Yes No
Article validation Yes No
Rate limiting (30 req/30s) Yes No
Retry logic for failures Yes No
Concepts API (ML topics) Yes No

While building devpub, I found several API endpoints that aren't documented anywhere obvious:

1. Semantic Search -- Dev.to has a full embedding-based search system using Gemini embeddings (768-dimensional vectors) with pgvector. You can search articles by meaning, not just keywords. The endpoint returns cosine similarity scores.

2. Concepts API -- These are ML-generated topic classifications with daily metrics: page views, reactions, comments, popularity scores. Way more powerful than manual tags.

3. V1 Accept Header -- The V1 API requires Accept: application/vnd.forem.api-v1+json

. Without it, you get V0 responses. I didn't see this mentioned in any competitor's code.

4. Nested analytics responses -- The analytics endpoints return nested objects like {"page_views": {"total": 246454, "average_read_time_in_seconds": 306}}

, not flat integers. Every tool I checked either doesn't use analytics or would break on this structure.

I built devpub's core in a day. Started at 2 PM on a Monday, had a working CLI by evening. Here's the honest timeline:

Hour 1-2: Research

Before writing a single line of code, I analyzed 9 competing tools. Downloaded them, read their source, mapped which API endpoints each one used. Found that the most "complete" tool covered 12 out of 40+ endpoints. Most covered 3-5.

Then I read the entire Forem API docs. Not the summary page that everyone reads. The full V1 spec. That's where I found semantic search, concepts, and the analytics endpoints that nobody knew existed.

Hour 3: Scaffolding

pyproject.toml

, src layout, Click CLI entry point. Boring stuff but I got devpub --help

working in 15 minutes. The key decision here: use httpx

over requests

. httpx gives you connection pooling, proper timeouts, and the async option for later without changing the interface.

Hour 4-5: The API client

This is where I spent the most time. Not because the HTTP calls are hard. Because I wanted the client to be production-grade from day one:

The first version didn't have any of this. It just called raise_for_status()

and threw ugly httpx.HTTPStatusError

exceptions at users. I caught that in testing when I pulled my own 86 articles and hit the rate limit at article 30. The whole thing crashed.

Hour 6: Testing against my real account

This is where things got interesting. My first devpub stats

call crashed with:

TypeError: '>=' not supported between instances of 'dict' and 'int'

Turns out the analytics endpoint returns {"page_views": {"total": 246454}}

, not {"page_views": 246454}

. Nested dicts. No existing tool handles this correctly because no existing tool uses analytics.

The health check endpoint also surprised me. In the V1 API (with the Accept header), /health_checks/app

requires authentication. Without the header, it returns 401. So I changed the health check to just call /users/me

instead.

Hour 7: The push --all scare

During testing, push --all

almost published my README.md to Dev.to. The original logic was: find any .md

file with a title

in frontmatter, push it. My README has YAML frontmatter with a title.

Fixed it by requiring both title

AND published

keys, and only scanning known directories (articles/

, posts/

, content/

, drafts/

). Small thing, but imagine accidentally publishing your CONTRIBUTING.md as a Dev.to article.

What I'd do differently

Start with the pull command, not push. Pull forces you to understand the API response format before you build the data model. I built the model first based on docs, then had to fix it when real responses looked different.

Mock tests from the start. I wrote all the code first, then tests. Should have written the API mock responses alongside the client methods. Would have caught the nested dict issue immediately.

Ship the rate limiter in v0.0.1. I initially thought "I'll add that later." Hit the limit within 30 minutes of real testing. Should have been there from commit one.

src/devpub/
  api/        # HTTP clients with rate limiting + retries
  cli/        # Click commands + Rich terminal output  
  core/       # Business logic (articles, sync, validation, config)
  templates/  # Article scaffolding (5 templates)

Key decisions:

$ pytest -v
57 passed in 1.08s

Tests mock the HTTP layer with respx

. No real API calls in CI. Covers:

pip install devpub
export DEVPUB_API_KEY=your_key_here
devpub doctor

Or from source:

git clone https://github.com/simplynadaf/devpub.git
cd devpub
pip install -e .

Get your API key at: https://dev.to/settings/extensions

Every command returns structured output, handles rate limits silently, and supports --dry-run

. AI coding agents (Claude Code, Copilot, Cursor) can use devpub as their publishing layer. The agent writes the article, devpub validates, pushes, and tracks performance. No human needed after the initial setup.

The project is in beta. PRs are welcome. Some things that need help:

--graph

flag is accepted but not implemented yetCheck the issues for good first issue

labels.

GitHub: github.com/simplynadaf/devpub

If this saves you time, star the repo. If something's broken, open an issue. If you want a feature, send a PR.

What's your current Dev.to workflow? Are you writing in the browser editor, or do you have a local setup? Curious what pain points people are hitting.

📺 Watch the full demo on YouTube

Built by Sarvar Nadaf - Cloud Architect

Follow me: Dev.to | GitHub | YouTube | LinkedIn

── more in #developer-tools 4 stories · sorted by recency
── more on @devpub 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/introducing-devpub-o…] indexed:0 read:6min 2026-08-01 ·