DocLogs

DocLogs is an engineer career operating system for capturing and reusing technical work.

What it solves

Engineers do valuable work every day: incident response, CI/CD and platform debugging, automation, infrastructure changes, PR reviews, and migrations. Most of that work is forgotten when it matters most: performance reviews, promotion packets, interviews, and professional storytelling.

DocLogs helps you turn day-to-day engineering activity into durable artifacts by:

• capturing evidence automatically from git, branches, PRs, and optional histories
• summarizing weekly progress and surfacing strong stories
• generating markdown artifacts for blog posts, LinkedIn, resumes, and interview prep
• keeping model usage provider-agnostic and safe with sanitization

Use --syntax on any command for usage, options, and examples (e.g. doclog generate --syntax). Standard --help is also available.

Core commands

• doclog capture
  • collect commits, repository activity, PR titles, tickets, and optional notes
  • store structured daily entries in local storage
• doclog weekly
  • review weekly work
  • surface candidate stories worth expanding
• doclog generate <type>
  • create reusable artifacts such as blog, linkedin, resume, interview, or changelog
  • default: saves a sanitized prompt file for manual use in Cursor or Copilot
  • optional: auto-generate via logged-in Cursor CLI or GitHub Copilot CLI
• doclog sanitize
  • sanitize captured content before any LLM request
  • redact internal URLs, tokens, IP addresses, and other sensitive details
• doclog publish
  • configure a git repo for publishing generated posts
  • commit and push a post to that repo

Recommended repository structure

./
├── README.md
├── config.yaml
├── commands/
│   ├── capture.py
│   └── config.py
├── docs/
│   ├── vision.md
│   └── architecture.md
├── doclog/
│   ├── entries/
│   ├── posts/
│   └── cache/
├── prompts/
│   ├── blog.md
│   ├── linkedin.md
│   └── resume.md

Architecture overview

DocLogs keeps the capture layer local and the generation layer model-agnostic. It is intentionally not a scheduler; use OS-level schedulers like cron, systemd --user, launchd, or Task Scheduler to invoke doclog capture at reminder times.

For implementation, start by focusing on:

1. evidence collection and local storage
2. weekly summaries and story selection
3. safe prompt generation
4. provider adapters for IDE CLIs (Cursor, Copilot) and optional API providers

Getting started

Install from PyPI

pip install -U doclogs-cli==0.1.10
doclog --help

Install from source (development)

git clone https://github.com/MridulTi/DocLogs.git
cd DocLogs
python -m venv .venv
source .venv/bin/activate
pip install -e .
doclog --help

First run

Data is stored under ~/.doclog/ (entries, posts, config). On first use, a default config.yaml is created there.

To keep using a project-local folder instead:

export DOCLOG_HOME="$PWD/doclog"
doclog capture

Daily use

1. Configure ~/.doclog/config.yaml (created automatically on first run).
2. Add a scheduler entry outside the CLI to invoke doclog capture at your preferred check-in time.
3. Capture daily progress and generate reusable career artifacts from the same captured story.

Generate posts

# Default: prompt file only (paste into Cursor/Copilot chat manually)
doclog weekly
doclog generate blog -t "nginx fix"

# Auto-generate via logged-in Cursor CLI
doclog config set provider cursor
doclog generate blog -t "nginx fix"

# Or Copilot
doclog config set provider copilot

# Back to prompt-only
doclog config set provider prompt_only

# Check provider availability
doclog config

Ensure Cursor/Copilot CLI is installed and authenticated when using those providers (agent login or copilot).

Publish a post to git

Point DocLogs at a local clone of your blog/docs repo, then commit and push generated posts. Each push copies the file under posts/YYYY-MM-DD/ (today's date) so re-publishing the same filename on a later day does not overwrite an earlier version.

doclog publish set repo ~/projects/my-blog
# or clone from GitHub automatically:
doclog publish set repo https://github.com/MridulTi/DocLogs
doclog publish set branch main
doclog generate blog -t 1
doclog publish push --latest

Or push a specific file:

doclog publish push ~/.doclog/posts/nginx-fix-blog.md -m "Add nginx TLS post"

Publish to PyPI (GitHub Actions)

Publishing runs via GitHub Actions — no local twine upload needed.

1. Configure trusted publishing on PyPI (leave Environment blank)
2. Push a version tag:

git tag v0.1.10
git push origin v0.1.10

The workflow uploads to PyPI automatically.

See docs/publishing.md for full setup.