solid-cli 1.1.0

A small SOLID-inspired CLI framework for easily adding commands.

solid-cli v1.1.0

A SOLID-inspired Python CLI toolkit for autonomous research, multimodal AI review (text / image / video / audio), and self-organising knowledge management. Built the way Uncle Bob would build it.

Quick Start

# Install in editable mode
python -m pip install -e .

# See all commands
python -m solid_cli --help

# Check version
solid-cli --version

# Agent onboarding — comprehensive guide for humans and AI agents
python -m solid_cli onboard
python -m solid_cli onboard --json   # machine-readable for agents

Commands

Command	What it does
research	Web-search via OpenAI → Markdown report in knowledge_base/
review	Upload file or text to Gemini → quality feedback in feedback/
download	Fetch sample image/video/audio for testing into samples/
read	Display a saved knowledge-base document
onboard	Full help guide for humans and AI agents
hello	Greeting example (--name Alice)
sum	Sum integers example (sum 1 2 3)

research — autonomous web search

solid-cli research "OpenAI Responses API best practices 2025" --tag openai_api
solid-cli research --list              # catalogue
solid-cli research --search apple      # filter by keyword
solid-cli research --batch queries.txt # bulk from file (tag|query format)
solid-cli research --stats             # aggregate library stats
solid-cli research --delete 3          # remove document #3
solid-cli research "topic" --force     # save even with 0 sources
solid-cli read 1                       # display document #1

Queries that return 0 sources are skipped by default to keep the library clean. Use --force to override.

review — multimodal Gemini quality review

solid-cli review photo.png "Check composition and quality"
solid-cli review clip.mp4  "Summarise content, assess quality"
solid-cli review song.wav  "Transcribe and assess audio quality"
solid-cli review code.py   "Code review for SOLID adherence"
solid-cli review --text "Some text with erors" "Check spelling"
solid-cli review --list              # catalogue
solid-cli review --read 1            # display feedback #1

download — fetch test assets

solid-cli download all               # image + video + audio
solid-cli download --url URL --name my_file.png
solid-cli download --list

Environment variables

Variable	Purpose
OPENAI_API_KEY	Required for research (web search via OpenAI)
GEMINI_API_KEY	Required for review (multimodal AI via Gemini)
KB_DIR	Override knowledge-base directory (default ./knowledge_base)
FEEDBACK_DIR	Override feedback directory (default ./feedback)

Place these in a .env file — python-dotenv loads it automatically.

Architecture (SOLID + Clean)

solid_cli/
├── command.py              ← Abstract Command interface (DIP)
├── registry.py             ← Auto-discovers commands (OCP)
├── cli.py / __main__.py    ← Thin entry points
├── services/
│   ├── search_service.py   ← SearchService ABC + OpenAI impl (SRP, LSP)
│   ├── review_service.py   ← ReviewService ABC + Gemini impl (SRP, LSP)
│   ├── storage_service.py  ← KnowledgeBaseStorage (SRP)
│   └── feedback_storage.py ← FeedbackStorage (SRP)
└── commands/
    ├── research.py         ← Web-search → knowledge_base/
    ├── review.py           ← Gemini multimodal → feedback/
    ├── download.py         ← Fetch test media → samples/
    ├── read.py             ← KB reader
    ├── onboard.py          ← Agent/human help system
    ├── hello.py            ← Example
    └── sum.py              ← Example

Output directories

knowledge_base/             ← research output
    index.json
    <date>/<topic>.md

feedback/                   ← Gemini review output
    index.json
    <date>/
        images/ video/ audio/ text/ documents/
            <review>.md

samples/                    ← downloaded test assets

Adding a new command

1. Create solid_cli/commands/my_cmd.py
2. Subclass Command and implement name, add_arguments, run
3. Done — the registry auto-discovers it, no wiring needed

External packages can also register commands via the solid_cli.commands entry-point group in their own pyproject.toml.