feat: add Higgsfield UI automation subagent for subscription credit usage#797
feat: add Higgsfield UI automation subagent for subscription credit usage#797marcusquinn merged 10 commits intomainfrom
Conversation
…sage Playwright-based browser automation that drives the Higgsfield web UI to generate images, videos, and apply effects using subscription credits (not available via API). Includes persistent auth, modal dismissal, site discovery, and full-quality PNG download via asset dialog. Tested end-to-end: prompt entry -> generation -> download (4x 1536x2048 PNGs).
|
Warning Rate limit exceeded
⌛ How to resolve this issue?After the wait time has elapsed, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout. Please see our FAQ for further information. WalkthroughAdds a new Higgsfield UI automation tool: a Bash CLI (higgsfield-helper.sh) that orchestrates Playwright-based browser automation, plus supporting package/config, route mappings, and documentation for UI-driven image/video/lipsync generation and pipeline workflows. (49 words) Changes
Sequence DiagramsequenceDiagram
participant User
participant CLI as higgsfield-helper.sh
participant Setup as setup()
participant Runtime as Node/Bun
participant Automator as playwright-automator.mjs
participant Browser as Higgsfield UI
User->>CLI: invoke command (e.g., image "prompt")
CLI->>Setup: ensure deps, Playwright, state, credentials
Setup-->>CLI: deps & auth OK / error
CLI->>Runtime: run `playwright-automator.mjs` via node/bun
Runtime->>Automator: execute automation script
Automator->>Browser: launch & navigate to routes
Automator->>Browser: interact (fill, click, submit)
Browser-->>Automator: page responses / assets
Automator-->>Runtime: save outputs / state
Runtime-->>CLI: exit status / results
CLI-->>User: print result or error
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRsPoem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing touches🧪 Generate unit tests (beta)
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 |
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 integrates a new subagent designed to automate interactions with the Higgsfield AI web interface. The primary goal is to enable the use of subscription credits for image and video generation, and access to UI-specific features like various apps and effects, which are not available through the existing API. This enhancement provides a robust and flexible way to leverage Higgsfield's full capabilities programmatically, ensuring seamless operation even with dynamic UI elements and offering high-quality output downloads. Highlights
Changelog
Activity
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
|
|
Review the following changes in direct dependencies. Learn more about Socket for GitHub.
|
🔍 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: Mon Feb 9 18:51:01 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
…e UI options - Rewrite generateVideo() with History tab polling and direct CDN download - Add downloadVideoFromHistory() for reliable video extraction - Add generateLipsync() for lipsync studio automation - Fix image generation detection: track queue items instead of image count - Fix start frame upload: use Promise.all with filechooser event - Fix credits pagination: set rows-per-page to 50 to show all 19 models - Add Unlimited mode switch toggle (ARIA role-based) - Add configureImageOptions() for aspect, quality, enhance, batch, preset - Expand model URL mapping to all 10 discovered image models - Add CLI options: --quality, --enhance, --sound, --batch, --unlimited, --preset - Update higgsfield-ui.md with complete capability map from UI exploration - Prompt input: try ARIA textbox, textarea, contenteditable (3 strategies) - Video completion: detect video src in newest History item, not just item count
🔍 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: Mon Feb 9 20:58:58 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
…ality and image detection - Add production pipeline: image -> video -> lipsync -> ffmpeg assembly - Add seed bracketing for systematic prompt testing (cost reduction) - Add Video Director subagent (compressed 14 tutorial PDFs into agent) - Fix video download: use dialog Download button for full-quality 1080p instead of CDN preview thumbnail from <video src> - Fix image generation detection: verify Generate click registered with retry logic, add 30s safety retry if no activity detected - Fix batch size control: use minus/plus counter pattern (— N/4 +) instead of non-existent numbered buttons - Update helper shell script with pipeline/seed-bracket/credits commands - Update higgsfield-ui.md with pipeline and video-director documentation
🔍 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: Mon Feb 9 22:48:29 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 7
🤖 Fix all issues with AI agents
In @.agents/scripts/higgsfield-helper.sh:
- Around line 96-100: Several shell functions (cmd_login, cmd_assets,
cmd_credits, cmd_screenshot, cmd_download, cmd_status) lack explicit returns on
their success path; update each function to include an explicit return (e.g.,
return 0) at the end of the successful execution path. Locate the bodies of
cmd_login, cmd_assets, cmd_credits, cmd_screenshot, cmd_download and cmd_status
and add a terminal "return 0" (or appropriate non-zero where failure is already
handled) so every execution path ends with an explicit return.
- Around line 52-59: The run_automator function currently picks the runner (node
or bun) and invokes "${AUTOMATOR}" but does not change directory, causing Node
module resolution (e.g., playwright) to fail; modify run_automator to cd into
HIGGSFIELD_DIR before executing the automator (use a subshell or pushd/popd to
avoid changing the caller's PWD), e.g., enter HIGGSFIELD_DIR, run "${runner}"
"${AUTOMATOR}" "$@", and then return to the original directory so module
resolution finds node_modules local to HIGGSFIELD_DIR.
- Around line 39-48: The Playwright presence check and installs should be
performed inside ${HIGGSFIELD_DIR} and failures must increment the missing
counter; update the node/bun checks to change into HIGGSFIELD_DIR (e.g. use node
-e "process.chdir('${HIGGSFIELD_DIR}'); require('playwright')" and bun -e
"process.chdir('${HIGGSFIELD_DIR}'); import 'playwright'") or wrap installs with
(cd "${HIGGSFIELD_DIR}" && ...), run bun/npm installs from that directory (use
bun install playwright || npm install playwright) without swallowing errors so
you can test their exit status, and if both install attempts fail increment the
missing variable (missing=$((missing+1))) before returning missing to ensure
check_deps reflects failure.
- Around line 62-76: In the setup() function the Playwright presence check and
install commands (the node -e "require('playwright')" check, bun install/npm
install playwright, and npx playwright install chromium) run in the caller's
current directory instead of in ${HIGGSFIELD_DIR}; change the block to cd into
${HIGGSFIELD_DIR} before running those commands (and restore the original
working directory afterwards or use a subshell) so installs and checks happen in
HIGGSFIELD_DIR, referencing setup(), the node -e "require('playwright')" check,
bun install playwright / npm install playwright, and npx playwright install
chromium to locate the code to modify.
In @.agents/scripts/higgsfield/routes.json:
- Around line 3-11: The route entry "seedream_4_5" in routes.json is
inconsistent with the other image routes and appears unused; either remove the
"seedream_4_5" key entirely from the JSON or correct its value to follow the
established pattern by changing its path to "/image/seedream-4-5"; after the
change, run a repo-wide search for "seedream_4_5" to confirm no code references
remain and update any callers to the corrected "/image/seedream-4-5" if needed.
In @.agents/tools/video/higgsfield-ui.md:
- Around line 58-65: Update the architecture note that currently says
"Playwright automation, ~2000 lines" to reflect the correct size or remove the
approximate line count: open the doc entry mentioning
higgsfield/playwright-automator.mjs (and the higgsfield-helper.sh wrapper) and
either change "~2000 lines" to the accurate "~1300+ lines" from the PR summary
or delete the line-count entirely so the description doesn't become stale.
In @.agents/tools/video/video-director.md:
- Around line 171-175: Add a blank line before the fenced code block that begins
with "```text" under the "Emotional Block Cues" heading so the block is preceded
by an empty line (fixing MD031). Locate the fenced block labeled ```text and
insert one blank line immediately above it to ensure the code block is properly
separated from the preceding paragraph/header.
🧹 Nitpick comments (2)
.agents/tools/video/higgsfield-ui.md (1)
80-84: Credentials stored as plaintext environment variables.The
credentials.shpattern withHIGGSFIELD_USERandHIGGSFIELD_PASSin plaintext is consistent with the existing framework convention. However, for a file containing a password, ensurecredentials.shhas restrictive permissions (chmod 600). Consider documenting this requirement..agents/scripts/higgsfield/package.json (1)
7-9: Consider pinning the Playwright dependency to improve build reproducibility in this DevOps automation context.The
^1.58.2caret range is standard npm practice, but for CI/CD automation where consistent browser behavior is critical, pinning to an exact version (1.58.2 — currently valid as of Feb 2026) eliminates risk of subtle breakage from patch-level Playwright updates.♻️ Optional: pin the version
"dependencies": { - "playwright": "^1.58.2" + "playwright": "1.58.2" }Since this is a private internal tool where reproducibility directly impacts automation reliability, pinning aligns with DevOps best practices for zero technical debt.
.agents/scripts/higgsfield-helper.sh
Outdated
| if ! node -e "require('playwright')" 2>/dev/null && ! bun -e "import 'playwright'" 2>/dev/null; then | ||
| print_warning "Playwright not found, installing..." | ||
| if command -v bun &>/dev/null; then | ||
| bun install playwright 2>/dev/null || npm install playwright 2>/dev/null | ||
| else | ||
| npm install playwright 2>/dev/null | ||
| fi | ||
| fi | ||
|
|
||
| return "${missing}" |
There was a problem hiding this comment.
Playwright install runs in the wrong working directory.
npm install playwright and bun install playwright execute in the caller's $PWD, not in ${HIGGSFIELD_DIR} where package.json lives. The installed node_modules will land in the wrong place, and the subsequent require('playwright') check (also not scoped to HIGGSFIELD_DIR) may pass or fail unpredictably.
Additionally, if both install commands fail (suppressed by 2>/dev/null), missing is never incremented, so check_deps returns 0 (success) even though Playwright is absent.
🔧 Proposed fix
- if ! node -e "require('playwright')" 2>/dev/null && ! bun -e "import 'playwright'" 2>/dev/null; then
+ if ! (cd "${HIGGSFIELD_DIR}" && node -e "require('playwright')" 2>/dev/null) \
+ && ! (cd "${HIGGSFIELD_DIR}" && bun -e "import 'playwright'" 2>/dev/null); then
print_warning "Playwright not found, installing..."
- if command -v bun &>/dev/null; then
- bun install playwright 2>/dev/null || npm install playwright 2>/dev/null
- else
- npm install playwright 2>/dev/null
- fi
+ (
+ cd "${HIGGSFIELD_DIR}" || exit 1
+ if command -v bun &>/dev/null; then
+ bun install || npm install
+ else
+ npm install
+ fi
+ ) || { print_error "Failed to install Playwright"; missing=1; }
fi
return "${missing}"🤖 Prompt for AI Agents
In @.agents/scripts/higgsfield-helper.sh around lines 39 - 48, The Playwright
presence check and installs should be performed inside ${HIGGSFIELD_DIR} and
failures must increment the missing counter; update the node/bun checks to
change into HIGGSFIELD_DIR (e.g. use node -e
"process.chdir('${HIGGSFIELD_DIR}'); require('playwright')" and bun -e
"process.chdir('${HIGGSFIELD_DIR}'); import 'playwright'") or wrap installs with
(cd "${HIGGSFIELD_DIR}" && ...), run bun/npm installs from that directory (use
bun install playwright || npm install playwright) without swallowing errors so
you can test their exit status, and if both install attempts fail increment the
missing variable (missing=$((missing+1))) before returning missing to ensure
check_deps reflects failure.
| "soul": "/image/soul", | ||
| "nano_banana": "/image/nano_banana", | ||
| "nano_banana_pro": "/image/nano_banana_2", | ||
| "seedream": "/image/seedream", | ||
| "seedream_4_5": "/seedream-4-5", | ||
| "wan2": "/image/wan2", | ||
| "gpt": "/image/gpt", | ||
| "kontext": "/image/kontext", | ||
| "nano_banana_inpaint": "/image/nano_banana?image-inpaint=true" |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
# First, verify the file exists and check the content around the flagged lines
if [ -f ".agents/scripts/higgsfield/routes.json" ]; then
echo "=== File exists, checking content ==="
head -15 .agents/scripts/higgsfield/routes.json | cat -n
echo ""
echo "=== Checking for seedream entries ==="
rg -n "seedream" .agents/scripts/higgsfield/routes.json
else
echo "File not found at .agents/scripts/higgsfield/routes.json"
fiRepository: marcusquinn/aidevops
Length of output: 770
🏁 Script executed:
# Search for references to seedream_4_5 or seedream-4-5 in the codebase
echo "=== Searching for seedream_4_5 route references ==="
rg -n "seedream.4.5|seedream-4-5" --type=js --type=json --type=py --type=ts
echo ""
echo "=== Also check for seedream route references in general ==="
rg -n "seedream" --type=js --type=ts --max-count=20Repository: marcusquinn/aidevops
Length of output: 254
🏁 Script executed:
# Check the structure of the routes.json to understand the pattern better
echo "=== Checking routes.json structure ==="
wc -l .agents/scripts/higgsfield/routes.json
echo ""
echo "=== Sample of image routes to confirm pattern ==="
rg -n "image" .agents/scripts/higgsfield/routes.json | head -15Repository: marcusquinn/aidevops
Length of output: 574
Remove or correct dead route seedream_4_5 — inconsistent path prefix and unused.
The route at line 7 breaks the established /image/ prefix pattern (maps to /seedream-4-5 instead of /image/seedream-4-5). More critically, no code in the repository references this route—searches across .js, .ts, .py, and .json files yield no hits. Either remove it if it's obsolete, or correct it to /image/seedream-4-5 if it should match the image routes convention. Leaving it as-is introduces configuration drift and technical debt.
🤖 Prompt for AI Agents
In @.agents/scripts/higgsfield/routes.json around lines 3 - 11, The route entry
"seedream_4_5" in routes.json is inconsistent with the other image routes and
appears unused; either remove the "seedream_4_5" key entirely from the JSON or
correct its value to follow the established pattern by changing its path to
"/image/seedream-4-5"; after the change, run a repo-wide search for
"seedream_4_5" to confirm no code references remain and update any callers to
the corrected "/image/seedream-4-5" if needed.
| ```text | ||
| higgsfield-helper.sh (shell wrapper) | ||
| └── higgsfield/playwright-automator.mjs (Playwright automation, ~2000 lines) | ||
| ├── Persistent auth state (~/.aidevops/.agent-workspace/work/higgsfield/auth-state.json) | ||
| ├── Site discovery cache (~/.aidevops/.agent-workspace/work/higgsfield/routes-cache.json) | ||
| ├── Credentials from ~/.config/aidevops/credentials.sh | ||
| └── Downloads to ~/Downloads/ (descriptive filenames: hf_{model}_{quality}_{prompt}_{ts}.ext) | ||
| ``` |
There was a problem hiding this comment.
Minor: Architecture diagram states ~2000 lines but PR summary mentions ~1300+ lines.
Keep the line count accurate or remove the approximation to avoid stale documentation. These numbers drift quickly.
🤖 Prompt for AI Agents
In @.agents/tools/video/higgsfield-ui.md around lines 58 - 65, Update the
architecture note that currently says "Playwright automation, ~2000 lines" to
reflect the correct size or remove the approximate line count: open the doc
entry mentioning higgsfield/playwright-automator.mjs (and the
higgsfield-helper.sh wrapper) and either change "~2000 lines" to the accurate
"~1300+ lines" from the PR summary or delete the line-count entirely so the
description doesn't become stale.
| **Emotional Block Cues** for dialogue: | ||
| ```text | ||
| "[Happy] Hello, [surprised] my [excited] name is Sarah!" | ||
| "[Concerned] Have you ever [frustrated] struggled with this?" | ||
| ``` |
There was a problem hiding this comment.
Missing blank line before fenced code block (MD031).
Per markdownlint, fenced code blocks should be surrounded by blank lines. There's a missing blank line before the code block on line 172.
♻️ Fix
**Emotional Block Cues** for dialogue:
+
```text
"[Happy] Hello, [surprised] my [excited] name is Sarah!"
"[Concerned] Have you ever [frustrated] struggled with this?"</details>
<details>
<summary>🧰 Tools</summary>
<details>
<summary>🪛 markdownlint-cli2 (0.20.0)</summary>
[warning] 172-172: Fenced code blocks should be surrounded by blank lines
(MD031, blanks-around-fences)
</details>
</details>
<details>
<summary>🤖 Prompt for AI Agents</summary>
In @.agents/tools/video/video-director.md around lines 171 - 175, Add a blank
line before the fenced code block that begins with "text" under the "Emotional Block Cues" heading so the block is preceded by an empty line (fixing MD031). Locate the fenced block labeled text and insert one blank line
immediately above it to ensure the code block is properly separated from the
preceding paragraph/header.
</details>
<!-- fingerprinting:phantom:poseidon:phoenix -->
<!-- This is an auto-generated comment by CodeRabbit -->
…PI-based download
- Model dropdown: filter buttons by x-position (<800) to exclude History sidebar matches
- Model name mapping: CLI names (kling-2.6) to UI text (Kling 2.6) with search fallback
- Unlimited mode: use getByRole('switch') for reliable ARIA matching
- Video detection: use item count + processing text instead of <video src> comparison
(all History items share the same CDN motion template URL)
- Video download: intercept fnf.higgsfield.ai/project API for CloudFront URLs (1080p)
with HTTP status validation and file size sanity checks
- Batch size: use getByRole('button', { name: 'Decrement', exact: true }) for inner button
- Start frame upload: 4-strategy approach (Upload button, Start frame area, hidden input, coordinates)
- Ensure output directory exists before download attempts
🔍 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: Tue Feb 10 01:06:58 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
…lient error handling - Video detection: use prompt matching instead of item count increase (History tab has fixed display limit ~12 items, count doesn't increase) - Image generation: initialize peakQueue from Phase 1 detection (fixes 300s timeout when queue resolves before first poll) - Image download fallback: limit to batch size instead of downloading all (batch=1 downloads 1 image, not all 24 on page) - Error handlers: wrap screenshot/browser.close in try/catch (prevents cascading errors when browser is already closed) - Video polling: refresh History tab when stuck, with progressive escalation (tab re-click at 30s, full page reload at 180s)
🔍 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: Tue Feb 10 02:27:36 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
…oll simultaneously - Add submitVideoJobOnPage(): submit a video job on an already-open page - Add pollAndDownloadVideos(): poll History for multiple prompts, download via API - Pipeline Step 3 now submits all scene videos first, then polls for all at once - Cuts N-scene video generation from N*4min to ~4min (3-scene test: 252s total) - Update higgsfield-ui.md with parallel pipeline docs and performance table
🔍 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: Tue Feb 10 03:00:37 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
…nterception - Add complete Remotion project (11 files) for animated captions, scene transitions, title cards, and multi-scene video assembly - Fix video download matching with exclusive jobSet claiming and character-by-character prefix scoring (min 20 chars) - Fix API interception in pollAndDownloadVideos: navigate to /create/video + click History tab instead of unreliable page.reload() - Add imagePrompts[] support for separate image vs video prompts - Fix missing fs imports (symlinkSync/unlinkSync -> copyFileSync) - Fix Remotion staticFile() usage (copies instead of symlinks) - Add calculateMetadata for dynamic duration/dimensions from props - Verified: 3-scene test (14.1s) and 4-scene Notion ad (18.6s) rendered
🔍 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: Tue Feb 10 05:10:06 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
|
Warning Review the following alerts detected in dependencies. According to your organization's Security Policy, it is recommended to resolve "Warn" alerts. Learn more about Socket for GitHub.
|
- Route nano-banana-pro to /nano-banana-pro feature page (has Unlimited switch) - Route seedream-4.5 to /seedream-4-5 feature page (has Unlimited switch) - Add Unlimited mode switch detection and toggle in generateImage() - Add direct API fetch fallback in pollAndDownloadVideos() and downloadVideoFromHistory() for when response interception misses - Verified: NBP generates 2 images in 20s, Seedream 4.5 in 5s, both unlimited - Verified: video download works with direct fetch fallback
🔍 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: Tue Feb 10 05:17:28 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
…ng, imagePrompts - Document Remotion post-production pipeline (captions, transitions, title cards) - Document imagePrompts[] for separate image vs video prompts - Update model table with correct unlimited routes (/nano-banana-pro, /seedream-4-5) - Document Unlimited mode switch on feature pages - Document direct API fetch fallback for video downloads - Add captions[], transitionStyle, transitionDuration to brief JSON format
🔍 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: Tue Feb 10 05:18:46 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
… line count - Fix npm install/require running in wrong directory (cd into HIGGSFIELD_DIR) - Fix run_automator to cd into HIGGSFIELD_DIR for correct module resolution - Add explicit return statements to all shell functions (cmd_login, cmd_assets, cmd_credits, cmd_screenshot, cmd_download, cmd_status) - Update architecture diagram line count from ~2000 to ~3400
🔍 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: Tue Feb 10 05:21:09 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
|
1 participant
Summary
Adds a complete Higgsfield UI automation subagent that drives the web UI via Playwright to generate images, videos, lipsync, and effects using subscription credits (not available through the API).
Key Features
Playwright Automator (
playwright-automator.mjs, ~3400 lines)Remotion Post-Production Pipeline (11 files)
Unlimited Model Routing
/nano-banana-profeature page (has Unlimited switch)/seedream-4-5feature page (has Unlimited switch)Video Download Reliability
fnf.higgsfield.ai/projectusing page auth cookiesBrief JSON with imagePrompts[]
imagePrompts[]for start frame generation vsscenes[].promptfor video animationcaptions[]for Remotion overlay with per-caption stylingtransitionStyleandtransitionDurationfor scene transitionsVerified E2E
Files Changed
.agents/scripts/higgsfield/playwright-automator.mjs— Main automator.agents/scripts/higgsfield/remotion/— 11 Remotion files (pipeline).agents/scripts/higgsfield-helper.sh— Shell wrapper.agents/tools/video/higgsfield-ui.md— Subagent documentation.agents/tools/video/higgsfield.md— API subagent reference