feat: add Video main agent for AI video generation and prompt engineering#161
feat: add Video main agent for AI video generation and prompt engineering#161marcusquinn merged 1 commit intomainfrom
Conversation
…ring Create dedicated video.md primary agent that owns the tools/video/ subagents (video-prompt-design, remotion, higgsfield). Auto-discovered by setup.sh and generate-opencode-agents.sh in alphabetical order. Includes subagent frontmatter for OpenCode Tab switching, integration points with content/social-media/marketing agents, and workflow documentation for prompt engineering, Remotion, and Higgsfield pipelines.
Summary of ChangesHello @marcusquinn, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly expands the system's multimedia content creation capabilities by introducing a dedicated 'Video' main agent. This new agent centralizes tools and workflows for AI video generation, programmatic video creation, and advanced prompt engineering, enabling users to efficiently produce high-quality video content with consistent characters and controlled audio. Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
WalkthroughThis pull request introduces a new Video agent to the framework documentation. It adds agent metadata, specifications, and routing entries across three files, establishing the Video agent as a primary orchestrator for video generation workflows with three specialized subagents. Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~5 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces a new 'Video' main agent for AI video generation. The changes include adding the agent's definition file and updating AGENTS.md and README.md to include this new agent. The new agent file is well-structured and detailed. My review focuses on ensuring consistency and clarity. I've noted a potential point of confusion regarding the agent's mode being set to subagent despite it being a 'Main Agent'. Additionally, there's an inconsistency in the README.md where the context7 MCP is not listed for the Video agent, even though it's included in its subagents and is listed for other similar main agents. The PR description also mentions no custom MCPs are needed, which contradicts the agent's definition.
| --- | ||
| name: video | ||
| description: Video creation and AI generation - prompt engineering, programmatic video, generative models, editing workflows | ||
| mode: subagent |
There was a problem hiding this comment.
The mode is set to subagent, but this file defines a 'Main Agent' as stated in the H1 title and the PR description. This is potentially confusing. If subagent is the correct mode for a main agent, consider adding a comment in the frontmatter to clarify this for future maintainability. If it's incorrect, it should be updated to the proper mode.
| | Research | `research.md` | Research and analysis tasks | context7, augment | | ||
| | Sales | `sales.md` | Sales operations and CRM | augment | | ||
| | SEO | `seo.md` | SEO optimization, Search Console, keyword research | gsc, ahrefs, dataforseo, serper, context7, augment | | ||
| | Video | `video.md` | AI video generation, prompt engineering, programmatic video | augment | |
There was a problem hiding this comment.
For consistency with other main agents and the subagents listed in .agent/video.md, you should also include context7 in the list of enabled MCPs. The video.md file explicitly lists context7 as a subagent for research purposes, and other main agents like Plan+, SEO, and Research also list context7 in this table.
| | Video | `video.md` | AI video generation, prompt engineering, programmatic video | augment | | |
| | Video | `video.md` | AI video generation, prompt engineering, programmatic video | augment, context7 | |
🔍 Code Quality Report�[0;35m[MONITOR]�[0m Code Review Monitoring Report �[0;34m[INFO]�[0m Latest Quality Status: �[0;34m[INFO]�[0m Recent monitoring activity: 📈 Current Quality Metrics
Generated on: Sat Jan 24 02:31:07 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
🤖 Augment PR SummarySummary: Adds a new
🤖 Was this summary useful? React with 👍 or 👎 |
![augmentcode[bot]](https://avatars.githubusercontent.com/in/1027498?s=60&v=4){kind=small}
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Fix all issues with AI agents
In @.agent/video.md:
- Around line 21-83: Add a short credential storage/security note to the Quick
Reference section of .agent/video.md: mention the canonical credential docs in
AGENTS.md and the local helper script setup-local-api-keys.sh, instruct users to
store API keys in ~/.config/aidevops/mcp-env.sh with 600 permissions and never
in working directories, and reference the related auth docs in higgsfield.md for
completeness so readers know where to securely place keys for Veo 3, Sora,
Kling, Seedance, Higgsfield and Remotion.
🧹 Nitpick comments (1)
.agent/video.md (1)
54-77: Consider progressive disclosure: move detailed workflows to subagents.The three workflow sections contain step-by-step implementation details that would be better suited for the respective subagent files (
video-prompt-design.md,remotion.md,higgsfield.md). The main agent file should provide high-level workflow descriptions and pointer references to the subagents for detailed guidance.This aligns with the framework's progressive disclosure pattern, keeping the main agent focused and reducing token usage until specific workflows are needed.
♻️ Recommended refactor to align with progressive disclosure pattern
## Workflows -### AI Video Prompt Engineering - -1. Define character with 15+ attributes for consistency -2. Structure prompt using 7 components (Subject, Action, Scene, Style, Dialogue, Sounds, Technical) -3. Include camera positioning syntax and negative prompts -4. Specify environment audio explicitly to prevent hallucinations -5. Keep dialogue to 12-15 words for 8-second generations - -### Programmatic Video (Remotion) - -1. Define compositions with `useCurrentFrame()` and `useVideoConfig()` -2. Drive all animations via `interpolate()` or `spring()` -3. Use `<Sequence>` for time-offset content -4. Render via CLI or Lambda for production - -### AI Generation Pipeline (Higgsfield) - -1. Generate base image with text-to-image (Soul, FLUX) -2. Create character for consistency across generations -3. Convert to video with image-to-video (DOP, Kling, Seedance) -4. Poll for completion via webhooks or status API +### AI Video Prompt Engineering + +Use the 7-component meta prompt framework for consistent AI video generation. See `@video-prompt-design` for character definition, camera positioning, audio design, and hallucination prevention strategies. + +### Programmatic Video (Remotion) + +Create videos programmatically using React components with frame-based animations. See `@remotion` for composition patterns, animation APIs, sequence management, and rendering workflows. + +### AI Generation Pipeline (Higgsfield) + +Access 100+ generative media models through a unified API. See `@higgsfield` for text-to-image, image-to-video workflows, character consistency, and webhook integration.This keeps the main agent file concise while directing users to the appropriate subagents for implementation details.
Based on coding guidelines: Use progressive disclosure pattern with pointers to subagents instead of inline content.
| # Video - Main Agent | ||
|
|
||
| <!-- AI-CONTEXT-START --> | ||
|
|
||
| ## Quick Reference | ||
|
|
||
| - **Purpose**: AI video generation and programmatic video creation | ||
| - **Subagents**: `tools/video/` (prompt design, Remotion, Higgsfield) | ||
|
|
||
| **Capabilities**: | ||
| - AI video prompt engineering (Veo 3, Sora, Kling, Seedance) | ||
| - Programmatic video creation with React (Remotion) | ||
| - Multi-model AI generation via unified API (Higgsfield) | ||
| - Character consistency across video series | ||
| - Audio design and hallucination prevention | ||
|
|
||
| **Typical Tasks**: | ||
| - Craft structured prompts for AI video generation | ||
| - Build programmatic video pipelines | ||
| - Generate consistent character series | ||
| - Design camera work, dialogue, and audio | ||
| - Compare and select AI video models | ||
|
|
||
| <!-- AI-CONTEXT-END --> | ||
|
|
||
| ## Subagent Reference | ||
|
|
||
| | Subagent | Purpose | | ||
| |----------|---------| | ||
| | `video-prompt-design` | 7-component meta prompt framework for Veo 3 and similar models | | ||
| | `remotion` | Programmatic video creation with React - animations, compositions, rendering | | ||
| | `higgsfield` | Unified API for 100+ generative media models (image, video, voice, audio) | | ||
|
|
||
| ## Workflows | ||
|
|
||
| ### AI Video Prompt Engineering | ||
|
|
||
| 1. Define character with 15+ attributes for consistency | ||
| 2. Structure prompt using 7 components (Subject, Action, Scene, Style, Dialogue, Sounds, Technical) | ||
| 3. Include camera positioning syntax and negative prompts | ||
| 4. Specify environment audio explicitly to prevent hallucinations | ||
| 5. Keep dialogue to 12-15 words for 8-second generations | ||
|
|
||
| ### Programmatic Video (Remotion) | ||
|
|
||
| 1. Define compositions with `useCurrentFrame()` and `useVideoConfig()` | ||
| 2. Drive all animations via `interpolate()` or `spring()` | ||
| 3. Use `<Sequence>` for time-offset content | ||
| 4. Render via CLI or Lambda for production | ||
|
|
||
| ### AI Generation Pipeline (Higgsfield) | ||
|
|
||
| 1. Generate base image with text-to-image (Soul, FLUX) | ||
| 2. Create character for consistency across generations | ||
| 3. Convert to video with image-to-video (DOP, Kling, Seedance) | ||
| 4. Poll for completion via webhooks or status API | ||
|
|
||
| ## Integration Points | ||
|
|
||
| - `content.md` - Script writing and content planning | ||
| - `social-media.md` - Platform-specific video formatting | ||
| - `marketing.md` - Campaign video production | ||
| - `seo.md` - Video SEO (titles, descriptions, thumbnails) |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
# Check the directory structure for video-related agents
find .agent -type f -name "*.md" | grep -i video | head -20Repository: marcusquinn/aidevops
Length of output: 898
🏁 Script executed:
# Search for credential-related content in video-related files
rg -i "api.{0,5}key|credential|auth|token|secret" .agent --glob "*.md" | grep -i videoRepository: marcusquinn/aidevops
Length of output: 4110
🏁 Script executed:
# Check main AGENTS.md for security protocols and credential guidance
cat -n .agent/AGENTS.md | head -100Repository: marcusquinn/aidevops
Length of output: 4167
🏁 Script executed:
# Look for the setup script mentioned in learnings
find . -name "setup-local-api-keys.sh" -o -path "*scripts/setup*" | head -20Repository: marcusquinn/aidevops
Length of output: 188
🏁 Script executed:
# Check the full content of video.md
cat -n .agent/video.md | head -150Repository: marcusquinn/aidevops
Length of output: 3234
🏁 Script executed:
# Check if higgsfield.md authentication section is comprehensive
cat -n .agent/tools/video/higgsfield.md | grep -A 20 "## Authentication"Repository: marcusquinn/aidevops
Length of output: 1350
🏁 Script executed:
# Verify the setup script exists and what it does
head -50 .agent/scripts/setup-local-api-keys.shRepository: marcusquinn/aidevops
Length of output: 1264
Add credential storage reference to video.md security context.
The video.md agent references multiple third-party services (Veo 3, Sora, Kling, Seedance, Higgsfield, Remotion) that require API keys. While higgsfield.md documents authentication, video.md should reference the main AGENTS.md credential storage protocols or the setup-local-api-keys.sh script in its Quick Reference section to ensure users understand where to securely store these keys (~/.config/aidevops/mcp-env.sh with 600 permissions, never in working directories).
🤖 Prompt for AI Agents
In @.agent/video.md around lines 21 - 83, Add a short credential
storage/security note to the Quick Reference section of .agent/video.md: mention
the canonical credential docs in AGENTS.md and the local helper script
setup-local-api-keys.sh, instruct users to store API keys in
~/.config/aidevops/mcp-env.sh with 600 permissions and never in working
directories, and reference the related auth docs in higgsfield.md for
completeness so readers know where to securely place keys for Veo 3, Sora,
Kling, Seedance, Higgsfield and Remotion.
1 participant
Summary
video.mdas a new primary agent (Tab-switchable in OpenCode)tools/video/subagents: video-prompt-design, remotion, higgsfieldgenerate-opencode-agents.shon nextsetup.shrunChanges
.agent/video.md- Primary agent with subagent frontmatter, workflows, integration points.agent/AGENTS.md- Added Video to progressive disclosure tableREADME.md- Added to primary agents table (15 total), alphabetical between SEO and WordPressArchitecture
DEFAULT_TOOLSfrom generate-opencode-agents.sh (write, edit, bash, read, glob, grep, webfetch, task, osgrep, augment)Summary by CodeRabbit
New Features
Documentation
✏️ Tip: You can customize this high-level summary in your review settings.