trtc-config-inspector

TRTC SDK configuration inspection and analysis tool. Downloads scene config template Excel, inspection result Excel, and project code from user-provided URLs, compares inspection results against target scene configurations, locates TRTC parameters in source code, and generates a structured modification report. The AI Agent then applies code changes based on the report. Supports live streaming, voice chat room, video call, and other scenarios. Triggers: TRTC config inspection, compare config and modify code, optimize TRTC scene configuration

Compares TRTC scene configuration templates against inspection results, generates a structured diff report, and guides the AI Agent or user to adjust source code accordingly.

Description

This skill implements TRTC SDK configuration inspection analysis and modification suggestion generation:

1. Download Resources: Download scene config template Excel, inspection result Excel, and project code archive from user-provided URLs
2. Parse & Compare: Parse both Excel files, compare current configuration against target configuration item by item
3. Locate Code: Find corresponding TRTC API call sites in the project source code
4. Generate Modification Plan: Output structured modification suggestions (JSON format) including target values, code locations, and change recommendations for each diff item
5. Output Report: Generate a diff summary table; the AI Agent then uses code editing tools to apply the changes

Note: The scripts themselves do not modify any user code. They only generate analysis reports and modification suggestions. Actual code changes are performed by the AI Agent using standard code editing tools (e.g., replace_in_file), and the user can review each modification step.

When To Use

Trigger this skill when the user makes requests like:

• "Help me modify the code based on this inspection result and live streaming config"
• "Download these files and optimize code according to the TRTC config template"
• "Compare TRTC inspection config and modify code per the template"
• "Check if TRTC SDK config meets the scene requirements"
• User provides scene config Excel (scene_config_.xlsx) and/or inspection result Excel (inspection_result_.xlsx)

How To Use

Step 1: Collect User Input

Confirm the following information with the user (if not fully provided):

Parameter	Description	Required
Scene Config Template	URL or local path to scene config Excel	Required
Inspection Result	URL or local path to inspection result Excel	Required (if absent, only output target config)
Code Download Link	URL to project code zip/tar.gz or local directory path	Required

Users may provide input in various ways:

• Direct URL links
• Local file/directory paths
• Files already present in the workspace

Step 2: Download Resource Files

Use the download script to fetch remote files:

python3 SKILL_BASE_DIR/scripts/download_files.py \
  --config-url "Scene config Excel URL" \
  --inspect-url "Inspection result Excel URL" \
  --code-url "Code archive URL" \
  --output-dir "SKILL_BASE_DIR/workspace"

This script will:

• Download scene config Excel to workspace/config.xlsx
• Download inspection result Excel to workspace/inspect.xlsx
• Download and extract code to workspace/code/
• Output JSON-formatted download results containing local paths for each file

If the user provides a local file path, skip the download for that file and use the local path directly.

Step 3: Parse and Compare Configuration

Run the diff script to generate a difference report:

python3 SKILL_BASE_DIR/scripts/diff_config.py "inspection_result_excel_path" "scene_config_excel_path"

The JSON report output contains:

• diffs — Config items where current value differs from target value (code modification needed)
• matches — Config items that already match (no modification needed)
• unable_to_compare — Items that cannot be automatically compared (manual review needed)

If only the scene config template is available (no inspection result), parse the template directly:

python3 SKILL_BASE_DIR/scripts/parse_excel.py "scene_config_excel_path" --type config

Step 4: Locate and Modify Code Based on Report

Based on the diffs array in the diff report, the AI Agent uses code editing tools to find corresponding API calls in the code and modify parameter values. Below are the mappings between config items and code APIs:

4.1 Room Entry Mode (Scene / AppScene)

Find enterRoom calls, modify the second parameter.

Config Value	Android Code Constant	iOS Code Constant
Live	TRTCCloudDef.TRTC_APP_SCENE_LIVE	TRTCAppSceneLIVE
VideoCall	TRTCCloudDef.TRTC_APP_SCENE_VIDEOCALL	TRTCAppSceneVideoCall
AudioCall	TRTCCloudDef.TRTC_APP_SCENE_AUDIOCALL	TRTCAppSceneAudioCall
VoiceChatRoom	TRTCCloudDef.TRTC_APP_SCENE_VOICE_CHATROOM	TRTCAppSceneVoiceChatRoom

4.2 Audio Quality

Find startLocalAudio calls.

Config Value	Android Code Constant	iOS Code Constant
MUSIC	TRTCCloudDef.TRTC_AUDIO_QUALITY_MUSIC	TRTCAudioQualityMusic
DEFAULT	TRTCCloudDef.TRTC_AUDIO_QUALITY_DEFAULT	TRTCAudioQualityDefault
SPEECH	TRTCCloudDef.TRTC_AUDIO_QUALITY_SPEECH	TRTCAudioQualitySpeech

4.3 System Volume Type

Find setSystemVolumeType calls.

Config Value	Android Code Constant	iOS Code Constant
media	TRTCCloudDef.TRTCSystemVolumeTypeMedia	TRTCSystemVolumeTypeMedia
voip	TRTCCloudDef.TRTCSystemVolumeTypeVOIP	TRTCSystemVolumeTypeVOIP
auto	TRTCCloudDef.TRTCSystemVolumeTypeAuto	TRTCSystemVolumeTypeAuto

4.4 Video Encoder Params

Find setVideoEncoderParam calls and TRTCVideoEncParam objects.

CRITICAL — Video encoder params must be compared item by item; do not miss any sub-parameter:

In the scene config Excel, video encoder params (resolution, FPS, bitrate) are split into multiple rows. The diff script outputs independent sub-parameter comparisons. All video encoder sub-parameter diffs must be checked and applied, including:

• Resolution: e.g., "recommended 720p" -> update videoResolution
• FPS: e.g., "25fps" -> update videoFps (frequently missed!)
• Bitrate: e.g., "720p-1800kbps" -> update videoBitrate

Parameter	Field Name	Description
Resolution	videoResolution	e.g., TRTC_VIDEO_RESOLUTION_1280_720 (720p)
FPS	videoFps	Integer, e.g., 15 or 25
Bitrate	videoBitrate	Integer (kbps), e.g., 1800
Min Bitrate	minVideoBitrate	Integer (kbps)
Resolution Mode	videoResolutionMode	PORTRAIT or LANDSCAPE
Adaptive Resolution	enableAdjustRes	Boolean

Resolution constant mapping:

• 540p -> TRTC_VIDEO_RESOLUTION_960_540
• 720p -> TRTC_VIDEO_RESOLUTION_1280_720
• 1080p -> TRTC_VIDEO_RESOLUTION_1920_1080

Recommended bitrates:

• 540p -> 1300 kbps
• 720p -> 1800 kbps
• 1080p -> 3000 kbps

4.5 Capture Volume / Remote Playback Volume

Find setAudioCaptureVolume and setAudioPlayoutVolume calls.

4.6 Experimental APIs

Find callExperimentalAPI calls with JSON string parameters. The Excel code_example column provides the specific JSON structure.

Step 5: Ensure Correct API Call Order

When the AI Agent modifies code, TRTC API calls must follow this order:

1. TRTCCloud.sharedInstance() — Create instance
2. setListener() — Set callback listener
3. setSystemVolumeType() — Set volume type (before enterRoom)
4. callExperimentalAPI() — Experimental settings (before enterRoom)
5. setVideoEncoderParam() — Set video encoder params
6. startLocalPreview() — Start local camera preview
7. startLocalAudio() — Start local audio capture
8. enterRoom() — Enter room

Step 6: Output Modification Report

After completing all modifications, output a summary table:

Config Item	Before	After	Status
Room Entry Mode	VideoCall	Live	Modified
Audio Quality	DEFAULT	MUSIC	Modified
System Volume Type	VOIP	Media	Modified
Video Resolution	640x480	1280x720	Modified
Video FPS	15fps	25fps	Modified
Video Bitrate	900kbps	1800kbps	Modified

For items marked as "optional", inform the user and let them decide whether to apply.

Implementation

Dependencies

• Python 3.8+
• openpyxl (auto-installed if missing)
• requests (auto-installed if missing)

Script Inventory

Script	Purpose
scripts/download_files.py	Download scene config, inspection result, and code archive from URLs and extract
scripts/parse_excel.py	Parse a single Excel file (scene config or inspection result) and output structured JSON
scripts/diff_config.py	Compare inspection result against scene config template and output diff report
scripts/run_inspector.py	End-to-end inspector entry point: download -> parse -> diff -> analyze code -> generate modification plan

Core Functions

• download_files.py::download_and_extract(config_url, inspect_url, code_url, output_dir) — Download and extract all resource files
• parse_excel.py::parse_config_excel(filepath) — Parse scene config Excel
• parse_excel.py::parse_inspect_excel(filepath) — Parse inspection result Excel
• diff_config.py::diff_configs(inspect_data, config_data) — Compare and output diffs

Edge Cases

• URL inaccessible: Prompt user to verify the link is valid or whether login authorization is required
• Abnormal Excel format: Catch parsing exceptions, prompt user to confirm the file format matches TRTC platform export standards
• API call not found in code: Mark as "new" in the report; the AI Agent adds it at the appropriate position following the API call order
• Optional config items: Items marked "optional" are not modified automatically; prompt user to decide
• Multi-platform code: Automatically detect code language (Kotlin/Java/Swift/ObjC) and adapt syntax accordingly
• Code archive format: Supports .zip and .tar.gz formats

Important Notes

• Excel files are exported by TRTC platform inspection tools and scene configuration tools
• When modifying code, preserve the existing code structure and only change specific parameter values
• For Kotlin projects, convert Java example code from Excel to Kotlin syntax
• The code_example column provides reference implementations; adapt to the project's actual code style
• Video encoder params must be checked item by item: The diff script outputs independent diffs for each sub-parameter (resolution, FPS, bitrate); all must be applied. FPS is the most commonly missed item

Security

• Scripts do not modify code: All Python scripts (download_files.py, parse_excel.py, diff_config.py, run_inspector.py) only perform download, parse, compare, and analyze operations. They do not directly modify any user source code files. Code modifications are performed by the AI Agent using standard editing tools, and the user can review each step
• URLs are user-controlled: Scripts only download files from URLs explicitly provided by the user; no hardcoded external addresses are accessed
• Extraction limited to workspace: Code archives are only extracted to the specified workspace output directory
• No credential storage: Scripts do not read, store, or transmit any keys, tokens, or user credentials
• Runtime dependencies: Only auto-installs openpyxl (Excel parsing) and requests (HTTP download), both standard PyPI packages