Skip to content

GH#14039: tighten video-higgsfield-ui.md (188→158 lines) #14049

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
alex-solovyev wants to merge 1 commit into
base: main
Choose a base branch
from
Open

GH#14039: tighten video-higgsfield-ui.md (188→158 lines) #14049

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 3 additions & 33 deletions .agents/content/video-higgsfield-ui.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -88,7 +88,7 @@ higgsfield-helper.sh pipeline --brief brief.json
| Nano Banana | `nano_banana` | 1cr | Yes |
| WAN 2.2 | `wan2` | 1cr | No |

Quality ranking (product shots): GPT → Seedream 4.5 → FLUX → Kontext → ... `--prefer-unlimited` (default) auto-selects best unlimited model via dedicated routes (e.g., `/nano-banana-pro`) — standard `/image/` routes cost credits even for subscribed models. Soul presets: All, New, TikTok Core, Instagram Aesthetics, Camera Presets, Beauty, Mood, Surreal, Graphic Art.
Quality ranking: GPT → Seedream 4.5 → FLUX → Kontext → ... `--prefer-unlimited` (default) selects best unlimited model via dedicated routes (`/nano-banana-pro`) — standard `/image/` routes cost credits even for subscribers. Soul presets: All, New, TikTok Core, Instagram Aesthetics, Camera Presets, Beauty, Mood, Surreal, Graphic Art.
Copy link
Copy Markdown
Contributor

@coderabbitai coderabbitai bot Mar 30, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🟡 Minor

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Description: Check if --prefer-unlimited flag exists in the higgsfield-helper.sh script

# Search for --prefer-unlimited flag definition and usage
rg -n --type=sh -C5 'prefer-unlimited' .agents/scripts/

# Search for route handling logic in the playwright automator
rg -n -C5 'nano-banana-pro|/image/' .agents/scripts/higgsfield/

Repository: marcusquinn/aidevops

Length of output: 16484

Clarify whether --prefer-unlimited describes CLI behavior or web UI routing logic.

The documentation describes real routing behavior (unlimited feature pages like /nano-banana-pro vs. standard /image/ routes that cost credits), but the framing is unclear. No --prefer-unlimited CLI flag was found in the scripts—the underlying mechanism is the web UI's model routing logic (IMAGE_MODEL_URL_MAP and UNLIMITED_MODELS in the Higgsfield automation code), not a CLI option.

Either clarify that this section describes web UI behavior specifically, or if --prefer-unlimited is intended as a documented CLI flag, verify it's exposed in the helper script's argument parsing and update the Commands examples section (lines 40–64) to show actual usage.

🤖 Prompt for AI Agents 
Verify each finding against the current code and only fix it if needed.

In @.agents/content/video-higgsfield-ui.md at line 91, Clarify that
`--prefer-unlimited` refers to web UI model-routing behavior (not a CLI flag) by
updating the text to state it uses the web routing maps `IMAGE_MODEL_URL_MAP`
and `UNLIMITED_MODELS`; alternatively, if you intended a CLI flag, implement and
expose `--prefer-unlimited` in the helper script's argument parsing and update
the Commands examples to show its actual usage. Locate references to
`--prefer-unlimited` in this doc, confirm the helper scripts for any existing
flag, and then either rephrase the doc to point to the UI routing logic
(`IMAGE_MODEL_URL_MAP`, `UNLIMITED_MODELS`) or add the flag implementation and
corresponding example commands.

Copy link
Copy Markdown
Collaborator Author

@alex-solovyev alex-solovyev Mar 30, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

--prefer-unlimited is a real CLI flag — defined in playwright-automator.mjs:221 and tested in higgsfield-commands.mjs:2182. The search missed it because it only searched .agents/scripts/ shell files, not the .mjs files in the higgsfield/ subdirectory. No change needed.

Copy link
Copy Markdown
Contributor

@coderabbitai coderabbitai bot Mar 30, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Verify --prefer-unlimited flag in the .mjs files
rg -n 'prefer-unlimited' .agents/scripts/higgsfield/ --type-add 'mjs:*.mjs' --type mjs -C3

# Also check higgsfield-commands.mjs around line 2182
rg -n 'prefer-unlimited' .agents/scripts/higgsfield/higgsfield-commands.mjs -C3

Length of output: 5638

@alex-solovyev, confirmed — my apologies for the noise. The verification shows --prefer-unlimited is fully implemented:

  • playwright-automator.mjs:221 — help text documents it as "Auto-select unlimited models when available (default: on)"
  • higgsfield-common.mjs:284,287 — flag parsed and mapped to preferUnlimited option
  • higgsfield-commands.mjs:2182–2188 — CLI flag parsing tested with assertions for both --prefer-unlimited and --no-prefer-unlimited

The original search missed it because it was scoped to --type=sh, excluding the .mjs files in .agents/scripts/higgsfield/. No change needed.

🐇 ears drooping slightly

(\_/)
( •_•)
/ >🔍  searched the wrong forest...
✏️ Learnings added 
Learnt from: alex-solovyev
Repo: marcusquinn/aidevops PR: 14049
File: .agents/content/video-higgsfield-ui.md:91-91
Timestamp: 2026-03-30T09:27:21.007Z
Learning: In marcusquinn/aidevops, CLI flags for the Higgsfield automation toolchain (e.g., `--prefer-unlimited`) are defined and parsed in `.agents/scripts/higgsfield/*.mjs` files (ES modules), not in shell scripts. When verifying CLI flag existence, always search `.mjs` files in `.agents/scripts/higgsfield/` using `rg` with `--type-add 'mjs:*.mjs' --type mjs`, in addition to `.sh` files. Key files: `playwright-automator.mjs` (help/docs), `higgsfield-common.mjs` (flag parsing), `higgsfield-commands.mjs` (tests).

Note: Learnings are effective only in the context of similar code segments. To apply general coding guidelines, please configure review instructions. You can manage existing learnings in the app.

🧠 Learnings used 
Learnt from: marcusquinn
Repo: marcusquinn/aidevops PR: 3840
File: .agents/tools/research/providers/wappalyzer.md:37-37
Timestamp: 2026-03-08T06:46:42.196Z
Learning: In marcusquinn/aidevops, the internal Node.js wrapper used by `.agents/scripts/wappalyzer-helper.sh` is `.agents/scripts/wappalyzer-detect.mjs` (ES module, `.mjs` extension — not `.js`). The helper script references it at lines 27 and 103. Any doc or review comment suggesting `.js` is incorrect.

Learnt from: alex-solovyev
Repo: marcusquinn/aidevops PR: 13484
File: .agents/services/communications/google-chat.md:18-18
Timestamp: 2026-03-30T00:27:14.196Z
Learning: In marcusquinn/aidevops, `google-chat-helper.sh` referenced in `.agents/services/communications/google-chat.md` is a planned/future helper script that is intentionally documented before its implementation exists in the repository. Do not flag its absence as a missing file or broken reference in reviews.

Learnt from: marcusquinn
Repo: marcusquinn/aidevops PR: 0
File: :0-0
Timestamp: 2026-03-28T10:23:27.439Z
Learning: In marcusquinn/aidevops, paths in `## Related` sections of `.agents/tools/git/` documents (e.g., `worktrunk.md`, `conflict-resolution.md`) use bare agent-readable doc pointers such as `workflows/git-workflow.md` — without filesystem-relative prefixes like `../../`. This is an intentional codebase convention for agent navigation, not broken markdown links. Do not flag these as incorrect relative paths.

Learnt from: marcusquinn
Repo: marcusquinn/aidevops PR: 0
File: :0-0
Timestamp: 2026-03-08T06:04:29.617Z
Learning: In runway-helper.sh, `--audio | -a` flags have two distinct behaviors: one path takes a boolean (true/false) validated by a case statement at lines 184-196, while other occurrences of `--audio | -a` accept URLs and do not require boolean validation.

Learnt from: marcusquinn
Repo: marcusquinn/aidevops PR: 3842
File: .agents/tools/content/summarize.md:0-0
Timestamp: 2026-03-08T20:13:11.476Z
Learning: In marcusquinn/aidevops, the correct Zhipu AI provider configuration for `.agents/tools/content/summarize.md` (steipete/summarize CLI) is: provider label `Z.AI (Zhipu)`, model ID `zai/glm-4.7`, and API key env var `Z_AI_API_KEY` (alias `ZAI_API_KEY`). The values `zhipu/glm-4` and `ZHIPU_API_KEY` are incorrect for this CLI. Fixed in commit 0e8e46b on PR `#3842`.

Learnt from: marcusquinn
Repo: marcusquinn/aidevops PR: 12005
File: .agents/tools/voice/cloud-voice-agents.md:143-146
Timestamp: 2026-03-28T16:34:14.801Z
Learning: In `.agents/tools/voice/cloud-voice-agents.md`, the `"<unit>"` string in MiniCPM-o 2.6 streaming_prefill content arrays (e.g., `{"role": "user", "content": ["<unit>", chunk]}`) is a valid and required TDM (time-division multiplexing) boundary marker in the official MiniCPM-o 2.6 streaming API. Do not flag it as incorrect or undocumented. The official repo uses `["<unit>", image, audio]` for multimodal chunks and `["<unit>", chunk]` for audio-only streaming.

Learnt from: CR
Repo: marcusquinn/aidevops PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-17T02:07:48.870Z
Learning: Applies to **/AGENTS.md : Limit root AGENTS.md to ~50-100 max instructions with universal applicability to >80% of tasks


### Video (5)

Expand Down Expand Up @@ -131,45 +131,15 @@ Wan 2.5 Fast, Kling 2.6 Lipsync, Google Veo 3, Veo 3 Fast, Wan 2.5 Speak (9cr),
}
```

`imagePrompts[]` = start-frame images. `captions[]` = Remotion overlay (styles: bold-white, minimal, impact, typewriter, highlight). Remotion setup: `cd .agents/scripts/higgsfield/remotion && npm install` — animated captions, transitions (fade/slide/wipe), title cards, 1080x1920.
`imagePrompts[]` = start-frame images. `captions[]` = Remotion overlay (styles: bold-white, minimal, impact, typewriter, highlight). Remotion: `cd .agents/scripts/higgsfield/remotion && npm install` — captions, transitions (fade/slide/wipe), title cards, 1080x1920.

## Output

`--headed` → `~/Downloads/higgsfield/` | headless → `~/.aidevops/.agent-workspace/work/higgsfield/output/` | override: `--output` | subdirs: `--project` | `--no-sidecar` `--no-dedup`

## CLI Options

| Flag | Description |
|------|-------------|
| `--model/-m` | Model slug |
| `--aspect/-a` | `16:9` `9:16` `1:1` `3:4` `4:3` `2:3` `3:2` |
| `--quality/-q` | `1K` `1.5K` `2K` `4K` |
| `--output/-o` | Output directory |
| `--headed/--headless` | Browser mode (headless default) |
| `--duration/-d` | Video seconds: 5, 10, 15 |
| `--image-file` | Input image path |
| `--image-file2` | Second image (multi-reference edit) |
| `--video-file/--motion-ref` | Motion reference video |
| `--batch/-b` | Images 1-4 |
| `--preset/-s` | Style preset name |
| `--seed` | Seed number |
| `--seed-range` | `1000-1010` or `"4000,4003,4008"` |
| `--brief` | Pipeline brief JSON path |
| `--tab` | `image` `video` |
| `--filter` | `image` `video` `lipsync` `upscaled` `liked` |
| `--asset-action` | `list` `download` `download-latest` `download-all` |
| `--asset-index` | 0-based asset index |
| `--chain-action` | `animate` `inpaint` `upscale` `relight` `angles` `shots` `ai-stylist` `skin-enhancer` `multishot` |
| `--feature` | `fashion-factory` `ugc-factory` `photodump-studio` `camera-controls` `effects` |
| `--subtype` | Vibe: `infographics` `text-animation` `posters` `presentation` `from-scratch` |
| `--project` | Organized output subdirs |
| `--prefer-unlimited` | Auto-select best unlimited model (default on) |

## Prompt Tips

- **Images** (camera + lighting + lens): `"Golden retriever, golden hour, shallow DOF, Canon EOS R5, 85mm, bokeh"`
- **Videos** (camera movement first): `"Smooth cinematic pan left to right, golden hour, 24fps film grain"`
- **Modifiers**: photorealistic → `"8k, highly detailed"` | cinematic → `"anamorphic, film grain, color graded"` | portrait → `"studio lighting, bokeh, 85mm"`
Images: camera + lighting + lens (`"Golden retriever, golden hour, shallow DOF, Canon EOS R5, 85mm, bokeh"`). Videos: camera movement first (`"Smooth cinematic pan left to right, golden hour, 24fps film grain"`). Modifiers: photorealistic `"8k, highly detailed"` | cinematic `"anamorphic, film grain, color graded"` | portrait `"studio lighting, bokeh, 85mm"`.

## Troubleshooting

Expand Down
Loading