AI Agents I Built (5/7) — Building an Automated Blogger API Publishing System

Designing a Markdown-Based Auto-Publish Pipeline with Blogger API v3 + OAuth 2.0

Key Summary

  • Blogger API v3 + OAuth 2.0 enables a pipeline that converts Markdown files into publishable HTML posts
  • blogger_publish.py handles the full flow — frontmatter parsing → body cleanup → HTML conversion → CSS injection → API post — in a single script
  • For bulk publishing, quota management (delay, retry, TOC suppression) is the critical design concern

Platform Selection Criteria

API access is mandatory for AI agent-driven content publishing. A comparison of platforms:

Platform API Cost AdSense Decision
WordPress.com REST API Paid plan required Paid only Cost overhead
Ghost REST API Self-hosted or paid Manual setup Ops overhead
Blogger v3 API Free Native integration Adopted

Blogger is free, exposes a REST API, and integrates natively with AdSense. For personal blog automation, this combination is the most practical.

Body

1. OAuth 2.0 Authentication Flow

Blogger API v3 requires OAuth 2.0 authentication. The full flow:

  1. Create OAuth 2.0 client ID in Google Cloud Console → download client_secret.json
  2. blogger_auth.py runs local server auth via InstalledAppFlow (port 8080)
  3. User logs into Google account in browser and grants permissions
  4. token.json is generated — contains access token + refresh token
  5. Subsequent requests auto-refresh using the refresh token
flow = InstalledAppFlow.from_client_secrets_file(client_secret, scopes)
creds = flow.run_local_server(port=8080)

Browser authentication is required only once. After that, the refresh token in token.json handles all renewals automatically — enabling the agent to publish without human intervention.

Token expiry warning: If the app is registered in "Testing" mode on Google Cloud Console, the refresh token is revoked after approximately 7 days. For long-term operation, the app must be switched to "Production" status to maintain a persistent refresh token.

2. blogger_publish.py — Core Publishing Script

This script is the center of the auto-publish pipeline. It accepts a Markdown file as input and publishes it to Blogger.

Processing flow:

Markdown file → frontmatter parse → body cleanup → HTML convert → CSS inject → API publish

Frontmatter parsing: Extracts title and labels from the --- block at the top of the Markdown file. Supports both Korean and English keys: ์ œ๋ชฉ:, ๋ผ๋ฒจ:, ํƒœ๊ทธ:, etc.

Body cleanup (clean_body): - H1 removal: Blogger renders the post title separately, making a Markdown H1 redundant. clean_body removes H1 to prevent double display. - Replaces ํ•ต์‹ฌ ์š”์•ฝ with "Key Summary" - Converts [Figure: description] notation into image placeholder divs

HTML conversion: Uses the Python markdown library with extra, tables, and fenced_code extensions.

CSS injection (BLOG_CSS): Injects a consistent style across all posts:

Element Style
Body text line-height 1.9, 16px
Code blocks Dark terminal style (background #1e1e2e)
Table headers Blue background (#4a90d9)
Blockquote Left blue border + light background, used for subtitles
Divider 2px solid #e0e0e0

Without direct CSS injection, Blogger's default theme styles apply to code blocks and tables, significantly reducing readability.

3. Command-Line Interface

python3 blogger_publish.py draft.md

python3 blogger_publish.py draft.md --draft

python3 blogger_publish.py draft.md --update POST_ID

python3 blogger_publish.py --list

python3 blogger_publish.py --delete POST_ID

python3 blogger_publish.py draft.md --blog en

The --blog kr/en flag switches between the Korean and English blogs. Each blog has a separate Blog ID; both share a single token.json for authentication.

4. Automatic TOC Update (update_toc.py)

Each time a post is published, the blog's table of contents page is updated automatically.

Operation: 1. Fetch full post list via Blogger API 2. Classify posts into categories by label (priority: Embedded > Claude Code > Local LLM > OpenClaw > AI Agents) 3. Generate HTML table of contents (by category + by series + by subgroup) 4. Update TOC page via Blogger Pages API

Series posts are grouped separately. OpenClaw posts are classified into subgroups: "Getting Started / Installation & Setup / Architecture & Advanced."

blogger_publish.py automatically calls update_toc.py after a successful publish. Use the --no-toc flag to suppress this behavior.

5. Bulk Publishing — Quota Management Design

Use batch_publish.py for bulk migration of existing posts or batch publishing.

for filepath in sorted_files:
    title, labels, html = parse_blog_post(filepath)
    publish_post(title, html, labels)
    time.sleep(10)  # 10-second delay between posts

Three quota design principles:

① 10-second delay between posts: The baseline interval to avoid Blogger API rate limits. Stays within the free quota reliably.

② 429 retry handling: On rate limit (HTTP 429), wait 60 seconds and retry. Without this logic, failures accumulate during bulk publishing.

③ TOC call suppression during batch operations: Per-post API call structure: - 1 publish call - TOC update: 1 full post list fetch + 1 page update

Updating the TOC on every publish triples the API call count. For bulk operations, suppress TOC updates with --no-toc and run it once after all posts are published — this is the quota-efficient design.

Bulk style update for existing posts: If CSS injection was added to the pipeline after posts were already published, those posts lack styling. batch_update_style.py injects CSS into all existing posts in bulk.

6. Dual-Language Operation (Korean + English)

Two blogs operated simultaneously:

Blog Purpose Flag
Korean blog Primary blog, Korean content --blog kr (default)
English blog English content, same topics --blog en

The same topic is written in both Korean and English and published to each respective blog. The script switches Blog IDs internally, so a single token.json handles authentication for both.

Conclusion

The core structure of the Blogger API auto-publish pipeline is three steps: Markdown → HTML → API. No complex CMS required — write in Markdown, publish with one script.

The most critical design consideration is quota management. Free APIs offer broad accessibility but impose daily call limits. For bulk operations, delay and retry logic are mandatory. Suppressing unnecessary chained API calls — such as updating the TOC on every single publish — is the key to stable long-term operation.

Series overview: Series index

๋Œ“๊ธ€

๋Œ“๊ธ€ ์“ฐ๊ธฐ

์ด ๋ธ”๋กœ๊ทธ์˜ ์ธ๊ธฐ ๊ฒŒ์‹œ๋ฌผ

Agent Memory Engine (2/10) — Building an AI Agent Memory System with SQLite Alone

← 1/10 SQLite Memory Engine f… ๐Ÿ“š Series Index 3/10 memcore Completed → A memory engine that runs without daemons, without dependencies, anywhere ํ•ต์‹ฌ ์š”์•ฝ This post covers how to design an AI agent memory engine (memcore) on a single-file SQLite backend. - Structure: 25 modules, ~6,464 lines, 22 DB tables - Principles: no daemon, single-file portability, no required external dependencies, host-agnostic - Core techniques: 4-layer hybrid search (topic · vector · FTS5 · LIKE), U-tag dialectic-based confidence evolution, ingestion-layer bias rejection rules What This Post Covers One approach to building AI agent memory without a server or cloud vector DB. Specifically: (1) the limitations of text-file-based memory, (2) vector DB alternatives compared and selection criteria, (3) how to layer search tiers, (4) how to evolve confidence scores over time, and (5) how to block biased memory accumulation at the ingestion stage. Design Principles: Why Single-File SQLite Text-file-based ...
Read more »

"ML Foundations (9/9) — PyTorch vs TensorFlow, and the Road to Local LLMs"

์ด๋ฏธ์ง€
← 8/9 Deep Learning Architec… ๐Ÿ“š Series Index (series end) The final part. So far we've focused on models. Now we focus on the tools that actually run them . We'll lay out the philosophical difference between PyTorch and TensorFlow, and trace how Hugging Face transformers , llama.cpp , MLX , and Ollama built the bridge to running large language models on your own machine. By the end you should have the full mental model of "download a pretrained LLM and serve it locally." 0. Learning Objectives Explain the eager-vs-graph difference between PyTorch and TensorFlow. Explain, in graph terms, how autograd automates the backward pass. Use Hugging Face transformers ' AutoModel , AutoTokenizer , and pipeline abstractions. Describe llama.cpp's GGUF quantization, INT4 inference, and CPU-first flow. Describe Apple MLX's use of unified memory and how it differs from PyTorch. Run a local LLM with Ollama and call it through an OpenAI-compatible API. ...
Read more »

"RAG Core Study (14/26) — Evaluation Sets with RAGAS & DeepEval"

์ด๋ฏธ์ง€
Series overview: Series index ๐Ÿ“š Series Index Without an evaluation set, every RAG improvement is a story, not evidence. RAG systems fail in more than one place: retrieval, context selection, answer generation, and grounding. That means "looks better" is not a metric. Part 14 explains how to build a golden dataset , how RAGAS and DeepEval fit into the workflow, and why teams must separate retrieval quality from answer quality if they want tuning decisions to hold up. 0. Prerequisites Part 13 reranking — system quality now depends on multi-stage ranking. Part 1 grounding — a correct-looking answer is not enough. Part 12 Hybrid — multiple retrieval settings need comparison on fixed data. 1. Learning Objectives Build a small but useful golden dataset for RAG. Distinguish retrieval metrics from answer metrics . Use RAGAS and DeepEval in the right roles. Avoid the common traps in judge-model-based evaluation. 2. ํ•ต์‹ฌ ์š”์•ฝ An evaluation set for RAG is ...
Read more »

"ML Foundations (8/9) — Deep Learning Architectures: CNN, RNN, Attention"

์ด๋ฏธ์ง€
← 7/9 Deep Learning Training ๐Ÿ“š Series Index 9/9 PyTorch vs TensorFlow → What did we keep adding on top of an MLP? Vision went CNN, sequences went RNN/LSTM, and eventually both converged on Attention and the Transformer. Each architecture is easier to remember if you read it as what it gave up to gain something else . This part is the one-line summary of the decisive inflection points: LeNet → AlexNet → ResNet → LSTM → Attention → Transformer. 0. Learning Objectives Explain why CNN's convolution, pooling, and weight sharing match images so well. Trace what got unblocked from LeNet → AlexNet → VGG → ResNet as depth grew. State BPTT for RNNs and the vanishing/exploding gradient problem. Write the LSTM gate equations (forget, input, output) with intuition. Write the attention formula in query/key/value form and the scaled dot-product variant. State the precise reasons Transformers replaced RNNs (parallelism + long-range dependencies). 1. ํ•ต์‹ฌ ์š”์•ฝ CNN : weight sharin...
Read more »

"ML Foundations (7/9) — Deep Learning Training: Optimizers, Regularization, Initialization"

← 6/9 Neural Networks ๐Ÿ“š Series Index 8/9 Deep Learning Architec… → Building an MLP, as we did in Part 6, is not the same as getting it to train well . The right optimizer, regularization, initialization, and learning rate — when these line up, deep networks converge. When they don't, the network refuses to learn at all. This part is the catalog of those crafts, with formulas, paper citations, and code in one place. 0. Learning Objectives Compare and write the update rules for SGD, Momentum, Nesterov, and Adam. Explain how Dropout, BatchNorm, and LayerNorm work and where they belong in a model. Derive the variance formulas for Xavier and He initialization and match them to activations. Implement step, cosine, and warmup learning-rate schedules in PyTorch. Explain why gradient clipping is effectively required for RNNs and Transformers. Diagnose the most common training failures (NaN, plateau, overfitting) and apply first-line fixes. 1. ํ•ต์‹ฌ ์š”์•ฝ SGD : \(w \leftarro...
Read more »

OpenClaw to Hermes Migration (2/13) — What to Preserve, Partially Port, or Discard

← 1/13 Current Structure Inve… ๐Ÿ“š Series Index 3/13 What Moves to Hermes → A code review record classifying operational scripts into three tiers: preserve, partial-preserve, and discard. What This Post Covers An asset classification framework (preserve / partial-preserve / discard) for migrating a legacy agent system to a successor A concrete case applying the Strangler Fig + Subset Migration strategy to a memory pipeline Classification results evaluated across LOC, external dependencies, and replaceability axes A method for slimming down a bloated single file (1,310 LOC) using the migration as a natural entry point Problem Definition: Port Everything or Cut Deep? The OpenClaw inventory comprises multiple agents, multiple scheduler/daemon processes, and multiple operational scripts. When moving to the successor system, Hermes, the choice is not binary. A full port carries accumulated technical debt along with it; a minimal port discards hard-won domain logic. Code-revie...
Read more »