@thunderkiller/video-clipper 1.1.0

package/.env.example

@@ -0,0 +1,130 @@
+# ── Provider selection ────────────────────────────────────────────────────────
+# Allowed values: openai | anthropic | google | xai | mistral | groq | zai | openrouter | custom
+# Default: openai
+LLM_PROVIDER=openai
+
+# ── API keys — set only the key for your chosen provider ─────────────────────
+OPENAI_API_KEY=your_key_here
+# ANTHROPIC_API_KEY=your_key_here
+# GOOGLE_GENERATIVE_AI_API_KEY=your_key_here
+# XAI_API_KEY=your_key_here
+# MISTRAL_API_KEY=your_key_here
+# GROQ_API_KEY=your_key_here
+# ZAI_API_KEY=your_key_here
+# OPENROUTER_API_KEY=your_key_here
+
+# ── Custom OpenAI-compatible provider ─────────────────────────────────────────
+# Set LLM_PROVIDER=custom and supply the base URL + API key for any endpoint
+# that speaks the OpenAI Chat Completions API (e.g. LM Studio, vLLM, Ollama,
+# LocalAI, Together AI, Fireworks AI, etc.).
+# Both vars are required when LLM_PROVIDER=custom.
+# CUSTOM_OPENAI_BASE_URL=http://localhost:11434/v1
+# CUSTOM_OPENAI_API_KEY=your_key_here
+
+# ── Model name — must match the chosen provider's model IDs ──────────────────
+# openai:    gpt-4o, gpt-4o-mini, gpt-4-turbo, ...
+# anthropic: claude-sonnet-4-5, claude-opus-4, claude-haiku-3-5, ...
+# google:    gemini-2.0-flash, gemini-1.5-pro, ...
+# xai:       grok-beta, grok-2, ...
+# mistral:   mistral-large-latest, mistral-small-latest, ...
+# groq:      llama-3.3-70b-versatile, llama-3.1-8b-instant, ...
+# zai:       glm-5, glm-4.7, glm-4.6, glm-5-turbo, ...
+# openrouter: meta-llama/llama-3.3-70b-instruct:free, google/gemma-3-27b-it:free, ...
+# custom:    depends on your endpoint (e.g. llama3.2, mistral, phi4, ...)
+LLM_MODEL=gpt-4o
+
+# ── Tunable parameters (all have defaults — only set if you want to override) ─
+# SCORE_THRESHOLD=7        # min score 1–10 to keep a segment
+# TOP_N_SEGMENTS=10        # max segments returned
+# CHUNK_LENGTH_SEC=120     # LLM chunk window size in seconds
+# CHUNK_OVERLAP_SEC=20     # overlap between consecutive chunks in seconds
+# MICRO_BLOCK_SEC=15       # micro-block grouping window in seconds
+# LLM_MAX_RETRIES=3        # max retries on rate-limit errors
+# DOWNLOAD_DIR=downloads/  # directory for yt-dlp downloads
+# OUTPUT_DIR=outputs/      # directory for clips and dumps
+# CACHE_DIR=outputs/cache  # directory for transcript and LLM result caching
+# LLM_CONCURRENCY=3        # max parallel LLM calls
+# CLIP_CONCURRENCY=1       # max parallel clip generation operations
+# DOWNLOAD_SECTIONS_MODE=all  # yt-dlp download mode: all (full video) or segments (individual clips)
+# FFMPEG_PRESET=fast       # ffmpeg encoding preset: ultrafast, superfast, veryfast, fast (default), medium, slow, slower
+# TIMESTAMP_OFFSET_SECONDS=0  # Adjust all clip timestamps (positive = later, negative = earlier) if transcript is misaligned with video
+
+# ── Custom system prompt ────────────────────────────────────────────────────────
+# Override the default LLM system prompt with a custom one. Useful for adapting
+# the analysis to specific content types (e.g., comedy, tech, education).
+# LLM_SYSTEM_PROMPT=You are an expert video editor specializing in viral clips.
+
+# ── LLM evaluation limits ─────────────────────────────────────────────────────
+# Cap the number of transcript chunks sent to the LLM. Useful for testing or
+# controlling API costs. Unset (default) means all chunks are evaluated.
+# MAX_CHUNKS=5
+
+# ── Output dumping ────────────────────────────────────────────────────────────
+# When true (default), writes two files after each run:
+#   OUTPUT_DIR/transcript/{videoId}.json  — raw normalized transcript lines
+#   OUTPUT_DIR/analysis/{videoId}.json    — full pipeline result (metadata + ranked segments)
+# Set to false to disable.
+# DUMP_OUTPUTS=true
+
+# ── Audio event detection ───────────────────────────────────────────────────────
+# Enable/disable audio event detection. When false, transcript-only mode is used.
+# Default: true
+# AUDIO_DETECTION_ENABLED=true
+
+# ── Transcript provider ─────────────────────────────────────────────────────────
+# Comma-separated ordered fallback chain. First provider to succeed wins.
+#   ytdlp   — yt-dlp auto-generated VTT subtitles (default, no audio required)
+#   whisper — local openai-whisper (Python), requires audio download + pip install openai-whisper
+#   gemini  — Gemini multimodal transcription (not yet implemented)
+# Default: ytdlp
+# TRANSCRIPT_PROVIDER=ytdlp
+
+# Audio provider:
+#   gemini  — Gemini Flash, semantic understanding, requires GOOGLE_GENERATIVE_AI_API_KEY
+#   yamnet  — local YAMNet (Python), class-ID based, requires: pip install tensorflow-hub soundfile numpy
+#   whisper — local openai-whisper (Python), speech-to-text + keyword match, requires: pip install openai-whisper
+#   both    — deprecated alias for "gemini,whisper"
+# Default: gemini,whisper
+# AUDIO_PROVIDER=gemini,whisper
+
+# Whisper model size to use when AUDIO_PROVIDER=whisper or AUDIO_PROVIDER=both.
+# Larger models are more accurate but slower. medium is the recommended default.
+# Options: tiny | base | small | medium | large-v3
+# Default: medium
+# AUDIO_WHISPER_MODEL=medium
+
+# Confidence threshold (0-1). Events below this value are discarded.
+# For Whisper: 1.0 = exact phrase match, 0.8 = partial keyword match.
+# For YAMNet: raw class score from the model.
+# Default: 0.30
+# AUDIO_CONFIDENCE_THRESHOLD=0.30
+
+# Pre-roll and post-roll for audio-only clip candidates (seconds before/after event)
+# Default: 5s pre-roll, 15s post-roll
+# AUDIO_CLIP_PRE_ROLL=5
+# AUDIO_CLIP_POST_ROLL=15
+
+# Time window (seconds) within which audio events boost LLM segment scores
+# Default: 10s
+# AUDIO_LLM_BOOST_WINDOW=10
+
+# Score boost applied to LLM segments when audio event is detected nearby
+# Default: 2
+# AUDIO_LLM_SCORE_BOOST=2
+
+# ── Game profile ────────────────────────────────────────────────────────────────
+# Game-specific event detection and LLM keywords:
+#   valorant:   gunshot, gunfire_burst, explosion + ace, clutch, defuse, spike
+#   fps:        gunshot, gunfire_burst, explosion + kill, streak, headshot
+#   boss_fight: explosion, crowd_cheering + phase, dead, finally
+#   general:    crowd_cheering, applause + insane, crazy, let's go
+# Default: general
+# GAME_PROFILE=general
+
+# ── Audio extra instructions ─────────────────────────────────────────────────
+# Optional extra instructions appended to the Gemini audio detection prompt,
+# after the game-profile preamble and before the JSON format block.
+# Use this to focus detection on specific moments, ignore certain sounds, or
+# add game-specific context not covered by the built-in profiles.
+# Example:
+# AUDIO_EXTRA_INSTRUCTIONS=Also flag moments where the caster's voice rises sharply. Ignore background music.

package/.github/workflows/ci.yml

@@ -0,0 +1,42 @@
+name: CI
+
+on:
+  push:
+    branches:
+      - main
+      - master
+  pull_request:
+    branches:
+      - main
+      - master
+jobs:
+  ci:
+    name: Type-check, Format, Test & Build
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Setup Node.js (LTS)
+        uses: actions/setup-node@v4
+        with:
+          node-version: '24'
+
+      - name: Setup pnpm
+        uses: pnpm/action-setup@v4
+
+      - name: Install dependencies
+        run: pnpm install --frozen-lockfile
+
+      - name: Type-check
+        run: pnpm type-check
+
+      - name: Format check
+        run: pnpm format:check
+
+      - name: Test
+        run: pnpm test
+
+      - name: Build
+        run: pnpm build

package/.github/workflows/release.yml

@@ -0,0 +1,72 @@
+name: Release
+
+on:
+  push:
+    branches:
+      - master
+
+jobs:
+  release:
+    name: Semantic Release
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      id-token: write
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          persist-credentials: false
+
+      - name: Setup pnpm
+        uses: pnpm/action-setup@v4
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '24'
+          registry-url: 'https://registry.npmjs.org'
+
+      - name: Setup npm auth
+        run: echo "//registry.npmjs.org/:_authToken=\${{ secrets.NPM_TOKEN }}" > ~/.npmrc
+
+      - name: Install dependencies
+        run: pnpm install --frozen-lockfile
+
+      - name: Build
+        run: pnpm build
+
+      - name: Configure Git
+        run: |
+          git config user.name "github-actions[bot]"
+          git config user.email "github-actions[bot]@users.noreply.github.com"
+          git remote set-url origin https://x-access-token:${{ secrets.PUSH_TOKEN }}@github.com/${{ github.repository }}.git
+
+      - name: Stage 1 GitHub Release (versioning, tags, releases)
+        id: github_release
+        run: npx semantic-release
+        env:
+          GITHUB_TOKEN: ${{ secrets.PUSH_TOKEN }}
+
+      - name: Verify GitHub Release
+        if: success()
+        run: |
+          echo "Checking for new git tags..."
+          CURRENT_TAG=$(git describe --tags --abbrev=0 2>/dev/null || echo "none")
+          echo "Current tag: $CURRENT_TAG"
+
+          if [ "$CURRENT_TAG" != "none" ]; then
+            echo "✅ New tag created: $CURRENT_TAG"
+            echo "Proceeding to npm publish..."
+          else
+            echo "❌ No new tag created - skipping npm publish"
+            exit 1
+          fi
+
+      - name: Stage 2 Publish to NPM
+        if: success()
+        run: npm publish
+        env:
+          NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}

package/.husky/pre-commit

@@ -0,0 +1,3 @@
+pnpm lint-staged
+pnpm type-check
+pnpm test

package/.prettierignore

@@ -0,0 +1,6 @@
+dist/
+node_modules/
+downloads/
+outputs/
+pnpm-lock.yaml
+package-lock.json

package/.prettierrc

@@ -0,0 +1,7 @@
+{
+  "singleQuote": true,
+  "semi": true,
+  "tabWidth": 2,
+  "trailingComma": "all",
+  "printWidth": 100
+}

package/.releaserc.json

@@ -0,0 +1,21 @@
+{
+  "branches": ["master"],
+  "plugins": [
+    "@semantic-release/commit-analyzer",
+    "@semantic-release/release-notes-generator",
+    [
+      "@semantic-release/changelog",
+      {
+        "changelogFile": "CHANGELOG.md"
+      }
+    ],
+    [
+      "@semantic-release/git",
+      {
+        "assets": ["package.json", "CHANGELOG.md"],
+        "message": "chore(release): ${nextRelease.version} [skip ci]"
+      }
+    ],
+    "@semantic-release/github"
+  ]
+}

package/AGENTS.md

@@ -0,0 +1,122 @@
+# Agent Instructions
+
+## Project
+
+TypeScript CLI that analyzes YouTube transcripts with an LLM to find interesting moments and optionally cut video clips.
+
+See `docs/plan.md` for the full architecture.
+
+## Stack
+
+- TypeScript (Node.js 18+)
+- Vercel AI SDK (`ai`, `@ai-sdk/openai`, `@ai-sdk/anthropic`, `@ai-sdk/google`, `@ai-sdk/xai`, `@ai-sdk/mistral`, `@ai-sdk/groq`) with `generateObject` + `zod` for structured LLM output
+- Multi-provider support: OpenAI, Anthropic, Google, XAI, Mistral, Groq, Zai, OpenRouter
+- `youtube-transcript` for transcript fetching
+- `yt-dlp` + `execa` for video download
+- `fluent-ffmpeg` for clip cutting
+- `zod` for config validation at startup
+- `p-limit` for concurrency control
+
+## Project Structure
+
+```
+src/
+  config/          # zod-validated env config — import this, never read process.env directly
+    index.ts
+    env.ts
+  services/        # core pipeline modules
+    urlParser/
+    metadataExtractor/
+    transcriptFetcher/
+    chunkBuilder/
+    llmAnalyzer/
+    segmentRanker/
+    clipRefiner/
+    videoDownloader/
+    clipGenerator/
+  types/           # shared TypeScript types
+    index.ts
+    config.ts
+    segment.ts
+    transcript.ts
+    video.ts
+    youtube-transcript.d.ts
+  utils/           # utility functions
+    cache.ts        # transcript + chunk LLM result caching
+    dumper.ts       # transcript/analysis JSON dumps
+    redactConfig.ts # config formatting for logs (redacts API keys)
+    format.ts       # timestamp formatting utilities
+    logger.ts       # logging utilities
+    modelFactory.ts # LLM provider factory
+  index.ts         # CLI entrypoint
+tests/             # all unit tests (mirrors module names)
+  urlParser.test.ts
+  chunkBuilder.test.ts
+  segmentRanker.test.ts
+downloads/         # yt-dlp output (gitignored)
+outputs/           # ffmpeg clip output, caches, dumps (gitignored)
+  cache/           # transcript + LLM result cache
+  transcript/       # transcript dumps
+  analysis/        # analysis dumps
+docs/
+  plan.md
+  free-models.md
+```
+
+## Code Rules
+
+- All code in TypeScript — no plain `.js` files
+- Every function must have explicit input/output types; avoid `any`
+- Use `zod` for all external data validation (LLM output, env vars, API responses)
+- Never read `process.env` directly — always import from `src/config.ts`
+- Never hardcode API keys, model names, thresholds, or directory paths — all come from config
+- Use `async/await` — no raw `.then()` chains
+- Use `Promise.allSettled` for parallel LLM calls so one failure doesn't abort the rest
+- Handle errors explicitly — no silent catches. Log a warning with the chunk index and reason on skip.
+
+## LLM Usage
+
+- Use `generateObject` (not `generateText`) for all LLM calls that return structured data
+- Define a `zod` schema for every LLM response before writing the prompt
+- Keep prompts in the same file as the function that uses them
+- Do not retry on malformed JSON — `generateObject` handles structured output natively
+- On any LLM call failure, catch and log, then continue — never crash the pipeline
+
+## Module Conventions
+
+Each module in `src/modules/` should:
+
+- Export a single main function named after the module (e.g. `fetchTranscript`, `buildChunks`)
+- Accept typed inputs and return typed outputs (no `any`)
+- Not import from other modules except `config.ts` and `types.ts` unless there is a clear dependency
+
+## Transcript Notes
+
+- `youtube-transcript` returns `offset` in **milliseconds** — normalize to seconds immediately after fetching
+- Micro-blocks group raw lines into ~15s windows before chunking
+- LLM chunks are 120s windows with 20s overlap — built from micro-blocks, not raw lines
+
+## Naming
+
+- Files: `camelCase.ts`
+- Functions: `camelCase`
+- Types/interfaces: `PascalCase`
+- Constants: `UPPER_SNAKE_CASE`
+- Zod schemas: `PascalCase` + `Schema` suffix (e.g. `SegmentSchema`)
+
+## Testing
+
+- Write unit tests for pure functions (URL parser, chunker, ranker, deduplicator)
+- Do not unit test functions that call external services (LLM, yt-dlp, ffmpeg) — integration test those separately
+- Test files live in `tests/` at the project root, mirroring the module name (e.g. `tests/urlParser.test.ts`)
+
+## Git
+
+- Commit messages: lowercase, imperative
+- One logical change per commit
+- Never commit `.env`, `downloads/`, or `outputs/`
+- It should be pointwise `-`
+- Shouldn't be more than 500 characters all together
+- it should be categorized as feat: , fix:, docs:, refactor etc...
+- also it should include specifics like feat(<feature>)
+- so final structure is <change ex: feat, docs, fix>: <short desc> followed by <full desc> pointwise

package/CHANGELOG.md

@@ -0,0 +1,45 @@
+# 1.0.0 (2026-03-18)
+
+### Bug Fixes
+
+- **audio:** align audio chunk windows with transcript LLM chunks ([f81981f](https://github.com/AmreetKumarkhuntia/video-clipper/commit/f81981fdca8c8b4b13682112551541b1df75a138))
+- **audio:** fix Gemini markdown fence parse error and add per-chunk caching ([3cc237f](https://github.com/AmreetKumarkhuntia/video-clipper/commit/3cc237f851c56236fb3f2795f5378b02cf34bde3))
+- **audio:** fix Gemini MM.SS timestamp ambiguity and make model configurable ([2139c5f](https://github.com/AmreetKumarkhuntia/video-clipper/commit/2139c5fa2030679bb460b402108f0e64482c0d0b))
+- **audio:** use python3 or python whichever available ([6501bf6](https://github.com/AmreetKumarkhuntia/video-clipper/commit/6501bf6ba99ba1364d285b7b921a22ecdfd23182))
+- github workflows ([18a9536](https://github.com/AmreetKumarkhuntia/video-clipper/commit/18a953619ed17de71d3c9bd0a86e1b42a10aea37))
+- **logging:** improve chunk analysis logs and gitignore outputs dir ([cc007de](https://github.com/AmreetKumarkhuntia/video-clipper/commit/cc007de0b1f453064a907689f9f30f6da2bebded))
+- **release:** add explicit npm auth setup step ([21c9287](https://github.com/AmreetKumarkhuntia/video-clipper/commit/21c928748fe90f278e67e34ae57553e5c279faf2))
+- **release:** add explicit npm auth setup step ([7e423fa](https://github.com/AmreetKumarkhuntia/video-clipper/commit/7e423fa76779d1dd6fe0d5bd0bac1cf0d54ffc8a))
+- **release:** remove github release plugin, keep npm only ([b37870b](https://github.com/AmreetKumarkhuntia/video-clipper/commit/b37870bfe997ad9d16ec1e0c22d52a07c99b4aaf))
+- **release:** restore github releases and fix git push permissions ([73e6fea](https://github.com/AmreetKumarkhuntia/video-clipper/commit/73e6fea88040844d00cebe79953f50f5ac96bc1e))
+- **release:** use PUSH_TOKEN instead of GITHUB_TOKEN ([d3414ad](https://github.com/AmreetKumarkhuntia/video-clipper/commit/d3414ad84944d48bc5dfcee7562b00c358bca27e))
+- **tests:** add global vitest setup to mock config module ([ecffd3b](https://github.com/AmreetKumarkhuntia/video-clipper/commit/ecffd3b940b4a2a3e5d0bd29911f2e70e22bdfe8))
+- yaml correction ([33c7854](https://github.com/AmreetKumarkhuntia/video-clipper/commit/33c7854015e1266b3fc01b6206da4ec946f94307))
+
+### Features
+
+- **analysis:** add per-chunk evaluations to analysis dump ([c562878](https://github.com/AmreetKumarkhuntia/video-clipper/commit/c562878971c65b8739b373cf2cac235d5c6697c0))
+- **audio:** add audio event result caching + requirements.txt ([5fd41d5](https://github.com/AmreetKumarkhuntia/video-clipper/commit/5fd41d5895e50060a93e9ef20621c85db52e80c4))
+- **audio:** add openai-whisper as local audio event detector ([12aa5be](https://github.com/AmreetKumarkhuntia/video-clipper/commit/12aa5be9a5521efef07e29f20ea93dd1a54aeda6))
+- **audio:** apply --max-parallel to audio event detection ([4996fc7](https://github.com/AmreetKumarkhuntia/video-clipper/commit/4996fc744f5000b791ecdd6677c95c24d5cf8e59))
+- **cache:** add chunk/transcript caching and format timestamps as HH:MM:SS ([4ea921d](https://github.com/AmreetKumarkhuntia/video-clipper/commit/4ea921d409c2b8cda2a1a5d59eb8bec14ca2c919))
+- **cache:** write chunk cache immediately after LLM analysis ([2ef3454](https://github.com/AmreetKumarkhuntia/video-clipper/commit/2ef3454a57c61abd5e0031639d6b79a726c7f413))
+- **ci:** add semantic-release for automated versioning ([0b7b6b6](https://github.com/AmreetKumarkhuntia/video-clipper/commit/0b7b6b65f685f486f3b51d21a78ac63108b87695))
+- **clip-generator:** add CLIP_CONCURRENCY config to prevent memory spikes ([be2c8af](https://github.com/AmreetKumarkhuntia/video-clipper/commit/be2c8af7763c5967b9b88d0b4406ac1a2c50af71))
+- **clip:** fix audio/video sync and add flexible download options ([c8ebe1a](https://github.com/AmreetKumarkhuntia/video-clipper/commit/c8ebe1a7f2218dcb881d10bd421e7b77b84009b6))
+- **llm:** add multi-provider support via modelFactory ([d72cb81](https://github.com/AmreetKumarkhuntia/video-clipper/commit/d72cb81d8adfe6366b60ef8ab74bbd7c38f67fe5))
+- **llm:** support custom system prompt via LLM_SYSTEM_PROMPT env var ([8b4525d](https://github.com/AmreetKumarkhuntia/video-clipper/commit/8b4525d9fffe2806c2232035b9a8553e56f37270))
+- **npm:** convert to scoped package @thunderkiller/video-clipper ([c8bf930](https://github.com/AmreetKumarkhuntia/video-clipper/commit/c8bf930c789cd9ceee3d8143fc5ff75674609f9a))
+- **output:** add transcript and analysis dump feature ([15db23c](https://github.com/AmreetKumarkhuntia/video-clipper/commit/15db23c77244b9db89604be463bccb97f6161987))
+- **phase4:** add video downloader and clip generator modules ([5c1cbde](https://github.com/AmreetKumarkhuntia/video-clipper/commit/5c1cbdeabe84a65013f7d5a8f34b3c2c336097e0))
+- **phase5:** add CLI flags, error handling, and progress logging ([36edf0a](https://github.com/AmreetKumarkhuntia/video-clipper/commit/36edf0a93da5389c7d5dee555fb6e3c7c7e47ee1))
+- **pipeline:** implement phase 2 core pipeline modules ([957b068](https://github.com/AmreetKumarkhuntia/video-clipper/commit/957b068a36dc29552c1a950ac74a8e337f097764))
+- **pipeline:** implement phase 3 LLM analysis and metadata modules ([8a088c9](https://github.com/AmreetKumarkhuntia/video-clipper/commit/8a088c99498824b35c9e348dd1ef8bb01eba7659))
+- **provider:** add custom OpenAI-compatible provider support ([5dc0c4e](https://github.com/AmreetKumarkhuntia/video-clipper/commit/5dc0c4e5fedddc97ba7907d5af89ba2fa46f2bc3))
+- **provider:** add openrouter support and free models doc ([e41d4a6](https://github.com/AmreetKumarkhuntia/video-clipper/commit/e41d4a6780dc5d72e83aafd1a1b4b8a46f2dd284))
+- **refiner:** add concurrency control and caching to segment refinement ([0738599](https://github.com/AmreetKumarkhuntia/video-clipper/commit/073859948700da0d59b806e29f03e9b7956f704d))
+- **release:** split into two sequential stages - GitHub then npm ([9f6d110](https://github.com/AmreetKumarkhuntia/video-clipper/commit/9f6d1108abd80cacf695d3872f4800fb84884bdd))
+- **scaffold:** bootstrap project structure and core infrastructure ([b187b3f](https://github.com/AmreetKumarkhuntia/video-clipper/commit/b187b3f43082ae88525f9444e84297b975167720))
+- **tooling:** add prettier, husky pre-commit hooks, and github actions ci ([fdbdc9a](https://github.com/AmreetKumarkhuntia/video-clipper/commit/fdbdc9a838e083d2777e6c6fb0f700b28fb1c324))
+- **version:** updated version ([69164f5](https://github.com/AmreetKumarkhuntia/video-clipper/commit/69164f59acca2e3236b4ab9a6e85441f89df01b4))
+- **yt-dlp:** add cookie support and improve error logging ([ae4afd4](https://github.com/AmreetKumarkhuntia/video-clipper/commit/ae4afd476b4176645bd69f40a62f22fd775356eb))