Agent-Skills.md

Agent Skills: AI Content Pipeline — Documentation Index

AI Content Pipeline documentation and architecture reference. Use when understanding codebase structure, AI models, CLI commands, YAML pipelines, provider APIs, error codes, or finding implementation details.

UncategorizedID: donghaozhang/video-agent-skill/project-docs

Repository

donghaozhang/video-agent-skill
donghaozhang
4318

Install this agent skill to your local

pnpm dlx add-skill https://github.com/donghaozhang/video-agent-skill/tree/HEAD/.claude/skills/project-docs

Skill Files

Browse the full folder contents for project-docs.

Download Skill

Loading file tree…

.claude/skills/project-docs/SKILL.md

Skill Metadata

Name
project-docs
Description
AI Content Pipeline documentation and architecture reference. Use when understanding codebase structure, AI models, CLI commands, YAML pipelines, provider APIs, error codes, or finding implementation details.

AI Content Pipeline — Documentation Index

73 AI models across 12 categories. Click-based CLI (aicp). Central model registry. YAML pipelines with parallel execution.

Key Facts

  • Package: video-ai-studio on PyPI, commands aicp / ai-content-pipeline
  • Version: 1.0.24, Python 3.10+
  • Registry: packages/core/ai_content_pipeline/ai_content_pipeline/registry.py + registry_data.py
  • CLI: Click framework in cli/click_app.py, commands in cli/commands/ (6 modules, 19 commands + vimax subgroup)
  • Providers: FAL AI (primary, 30+ models), Google (Gemini/Veo), ElevenLabs (TTS), OpenRouter (prompts), Replicate
  • API keys: FAL_KEY, GEMINI_API_KEY, ELEVENLABS_API_KEY, OPENROUTER_API_KEY in .env

Documentation Files

Load the specific file when Claude needs deeper information on that topic.

Models & Providers

| File | Load When | |------|-----------| | models.md | Selecting models, checking pricing, comparing options across 12 categories | | provider-comparison.md | Choosing between FAL AI, Google, ElevenLabs, OpenRouter, Replicate |

CLI & API

| File | Load When | |------|-----------| | cli-commands.md | CLI usage, command flags, global options (--json, --quiet, --stream, --input) | | python-api.md | Python API: AIPipelineManager methods, data classes, error handling | | aicp-vimax-commands.md | ViMax subgroup: novel2movie, idea2video, script2video pipelines |

Architecture & Code

| File | Load When | |------|-----------| | architecture.md | System design, data flow diagrams, component responsibilities | | package-structure.md | File locations, import paths, module dependencies |

Pipelines

| File | Load When | |------|-----------| | yaml-pipelines.md | YAML config syntax, 10 step types, variable interpolation, dependencies | | parallel-execution.md | Parallel groups, performance optimization, 2-3x speedup patterns |

Content Creation

| File | Load When | |------|-----------| | prompting.md | Writing effective prompts, templates, model-specific tips | | video-tips.md | Image-to-video vs text-to-video, motion prompts, model selection | | video-analysis.md | AI video analysis with Gemini, timeline/describe/transcribe modes |

Optimization

| File | Load When | |------|-----------| | cost-management.md | Pricing tables, budget strategies, cost estimation CLI/API | | performance.md | Speed benchmarks, batching, caching, network optimization | | best-practices.md | Project organization, pipeline design patterns, QA workflows |

Troubleshooting

| File | Load When | |------|-----------| | error-codes.md | Error codes AUTH/CFG/MDL/PIP/NET/FILE/RATE/COST/VAL with solutions | | troubleshooting.md | Diagnostic steps, common issues, installation/API/network problems | | faq.md | Frequently asked questions across 9 categories |

Development

| File | Load When | |------|-----------| | testing.md | Test strategies, mocks, fixtures, CI/CD, running pytest | | security.md | API key security, input validation, production deployment | | contributing.md | Development workflow, coding standards, adding models/providers | | migration.md | Version upgrades, breaking changes, rollback instructions |

Setup & Learning

| File | Load When | |------|-----------| | setup.md | Installation (PyPI/source), venv, API keys, first pipeline | | learning-path.md | Structured learning tracks: Quick Start, Comprehensive, Developer |

Examples

| File | Load When | |------|-----------| | basic-examples.md | Simple image/video generation, CLI examples | | advanced-pipelines.md | Multi-stage production, A/B testing, batch processing | | use-cases.md | Marketing, education, enterprise, gaming applications | | integrations.md | Flask, FastAPI, Celery, webhook patterns |

Model Categories Quick Reference

| Category | Count | Key Models | Cost Range | |----------|-------|------------|------------| | Text-to-Image | 8 | flux_dev, flux_schnell, imagen4, nano_banana_pro | $0.001-0.08 | | Image-to-Image | 8 | photon, kontext, clarity, seededit | $0.015-0.05 | | Text-to-Video | 10 | veo3, kling_3_pro, sora_2, hailuo_pro | $0.08-6.00 | | Image-to-Video | 15 | veo_3_1_fast, kling_3_pro_i2v, sora_2_i2v | $0.08-3.60 | | Video-to-Video | 4 | kling_o3_pro_edit, kling_o3_standard_edit | $0.25-0.34/s | | Avatar | 10 | omnihuman_v1_5, fabric_1_0, multitalk | $0.06-0.25/s | | Image Understanding | 7 | gemini_describe, gemini_detailed, gemini_qa | $0.001-0.002 | | Prompt Generation | 5 | openrouter_video_prompt + style variants | $0.002 | | Text-to-Speech | 3 | elevenlabs, elevenlabs_turbo, elevenlabs_v3 | $0.03-0.08 | | Speech-to-Text | 1 | scribe_v2 | $0.008/min | | Add Audio | 1 | thinksound | $0.001/s | | Upscale Video | 1 | topaz | ~$1.50/video |

CLI Commands Quick Reference

# Core generation
aicp generate-image --text "prompt" --model flux_dev
aicp create-video --text "prompt" --video-model kling_3_pro
aicp run-chain --config pipeline.yaml [--parallel] [--stream] [--dry-run]

# Media operations
aicp generate-avatar --image-url URL --audio-url URL --model omnihuman_v1_5
aicp analyze-video -i video.mp4 [-t timeline|describe|transcribe]
aicp transcribe --input audio.mp3 [--srt] [--raw-json]
aicp transfer-motion --image-url URL --video-url URL
aicp upscale-image --image photo.png --upscale 2
aicp generate-grid --text "prompt" --layout 2x2

# Discovery
aicp list-models [--category X] [--provider X] [--json]
aicp list-avatar-models | list-video-models | list-motion-models | list-speech-models

# Project
aicp setup | init-project | organize-project | structure-info | create-examples

# ViMax (novel-to-video)
aicp vimax idea2video --idea "concept"
aicp vimax novel2movie --novel novel.txt [--storyboard-only]
aicp vimax script2video --script story.txt

Architecture Summary

CLI (Click) → Pipeline Manager → Providers → External APIs
                    ↓                            ↑
              Config (YAML)              FAL / Google / ElevenLabs / OpenRouter
                    ↓
              Executor → Parallel Engine → Results → Output Files
  • Central Registry: registry.py defines ModelDefinition + ModelRegistry, registry_data.py registers all 73 models
  • Auto-Discovery: Generator classes use MODEL_KEY class attributes
  • CLI: Root Click group in cli/click_app.py, commands auto-registered at import time
  • Pipeline: manager.py orchestrates, executor.py runs steps, parallel.py handles concurrency

Testing Quick Reference

python -m pytest tests/ -v                    # Full suite (~844 tests)
python -m pytest tests/test_registry.py -v    # Registry tests
python -m pytest tests/test_click_app.py -v   # CLI tests
python scripts/validate_registry.py           # Registry validation
AI Content Pipeline — Documentation Index Skill | Agent Skills