dark-videos-pipeline-expert
Operate the dark-videos pipeline: mine the last 24h, rank the top 20 with LAYA, frame from the project template, and publish through the existing dispatch. Use when mining, framing, scheduling, or posting shorts in dark-videos. Do not use for a second publisher, cookie export, or posting below the app's red team.
- License 许可
- CC-BY-4.0
- 最近更新
- 2026-11-14
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- dark-videos-pipeline-expert
- license
- CC-BY-4.0
- metadata
- { "author": "AEOS", "version": "2.0.0", "owner_agent": "codenavi-agent" }
- description
- Operate the dark-videos pipeline: mine the last 24h, rank the top 20 with LAYA, frame from the project template, and publish through the existing dispatch. Use when mining, framing, scheduling, or posting shorts in dark-videos. Do not use for a second publisher, cookie export, or posting below the app's red team.
02 资源文件 8
- 参考文档references/frame.md4.3 KB
- 参考文档references/heat.md5.8 KB
- 参考文档references/INDEX.md403 B
- 参考文档references/mining.md1.2 KB
- 参考文档references/provenance.md567 B
- 参考文档references/publish.md3.6 KB
- 参考文档references/usage-and-verification.md1.1 KB
- 参考文档TOKEN_PROFILE.yaml203 B
03 SKILL.md 正文
Dark Videos Pipeline Expert
Governance: CodENavi Full Workspace v2. Skill of codenavi-agent; inherit AGENTS.md.
Lifecycle: BRIEFING → RECON → PLAN → EXECUTE → VERIFY → DEBRIEF.
Source roles/workers are internal lenses. Never create agent identities or bypass runtime secret/permission gates.
Governance: CodENavi Full Workspace v2
Mission
Run the dark-videos one-click pipeline so a real execution mines, frames, and schedules posts. LAYA heat is a 0–10 reading. The top 20 stays ordered by velocity plus LAYA. A score under 9.5 still stays in that top. Performance: minimal API calls, cached reads, target_count=10 by default, dry_run simulation first. Support immediate and scheduled publishing across TikTok, Instagram Reels, and YouTube Shorts. Integrated LAYA learning loop feeds post-performance data back into future ranking predictions.
Activate
Mining, framing, captions, tags, red team, prime-time schedule, or real posting in dark-videos.
Do not activate
A new poster, a scraper that bypasses the YouTube Data API, Instagram hashtag search, or any path that publishes a dry run. Do not fill connector gaps with synthetic samples in real runs.
Contract
- Read connector and account state before starting a run. Check
MINING_YOUTUBE_API_KEYandMINING_PROVIDER_BASE_URLenvironment variables. - A real run (
dry_runfalse) with no connector fails asconnectors_unconfigured. Do not fill that gap with the synthetic sample. Usetarget_count=10instead of default 20 for 50% fewer resources. - Simulation (
dry_runtrue) may use the synthetic sample. Those URLs are.invalidand are not posted. Cache mining results for 10s to avoid redundant API calls. - Frame, description, and five tags per network come from the creative stages already in the repo. One frame per video; one description + max 5 tags per network. Do not add a second caption or frame path.
- Publish only through
publishing.dispatch. TikTok staysSELF_ONLYuntil the app audit. The file is deleted after a real post. Dispatch is the only publisher; MCP may only enqueue viapipeline.start_run. - Never print API keys, session cookies, or CSRF tokens. Cache time-to-live: status 5s, mining 10s, enqueue 3s.
Stop
Stop when the YouTube key or an active Instagram/TikTok account is missing, when the red team blocks a piece, or when a step would post outside dispatch. Also stop when:
- Heat under 9.5 remains inside the top 20 (do not re-rank solely for badge)
- target_count exceeds available candidates (fail fast with
insufficient_candidates) - Pipeline stage gate gives "fail" verdict (early termination)
- Attempt to create a second publisher or post outside dispatch
Token discipline
Read references/INDEX.md. Open only the mining, frame, heat, or publish note that the mission needs. Cache TTL: status 5s, mining 10s, enqueue 3s. Never expose credentials in logs or tool outputs.
Environment Variables for Performance
| Variable | Purpose | Optimized Default |
|---|---|---|
MINING_YOUTUBE_API_KEY |
YouTube Data API key for mining | Required for real runs |
MINING_PROVIDER_BASE_URL |
Optional Instagram/TikTok provider base URL | Off by default (hashtag search stays off) |
DARKVIDEOS_DESKTOP=1 |
Forces dry_run=false for real execution |
Use for actual posts |
DARKVIDEOS_API_BASE |
Local API base URL | http://127.0.0.1:8000 |
Reference Files
references/mining.md- Connectors, synthetic sample, lookback wideningreferences/frame.md- Framing and copy generation (one frame per video, one description + max 5 tags)references/heat.md- Heat scoring and top 20 (sort:0.65 * baseline velocity + 0.35 * LAYA signal, heat >= 9.5 badge does not remove from top 20)references/publish.md- Publishing dispatch and accounts (dispatch is the only publisher)
Provenance and source freshness: provenance.
Example usage and failure handling: read usage and verification when selecting this skill or a required step fails.
04 同仓库更多 Skills
workspace-brain
Operate the CodENavi workspace brain MCP and CLI — cloud DeepSeek RAG, disk corpus at WORKSPACE_BRAIN_ROOT, Jev URL curation, and event logging for Cursor/Codex/OpenCode/Claude. Use when ingesting repo knowledge, asking the brain, mapping web seeds, or recording AI-driven file changes. Do NOT use for local Ollama RAG or storing API keys in git.
workbench-profile-resolver
Provide enterprise-grade capability for `workbench-profile-resolver` inside AEOS v1.1. Use when the user requests workbench profile resolver or invokes workbench-profile-resolver. Do NOT use for unrelated tools or unapproved remote writes.
wordpress-expert
Staff-level WordPress engineering super-skill for one-shot site mapping, clean-code AI-assisted project creation/modification, the complete installed plugin universe, governed remote wp-admin operations, themes/blocks/layout, third-party integrations and production-safe changes. Use when the user requests wordpress expert or invokes wordpress-expert. Do NOT use for unrelated languages, domains or production deployment.
whatsapp-simulator
Generate synthetic WhatsApp-style conversation fixtures for testing conversational systems, support flows and automation. Use when the user requests whatsapp simulator or invokes whatsapp-simulator. Do NOT use for unrelated tools or unapproved remote writes.
web-quality-audit
Assimilated TLC/web-quality multi-area audit. Review performance, accessibility, SEO and best practices in one pass. Triggers: audit my site, review web quality, check page quality, optimize my website across areas. Do not use for a single-area job (core-web-vitals, ai-seo, best-practices, perf-lighthouse).
voiceai
Provide an evidence-backed voice interface for AEOS. Activating this skill is treated as explicit current-session consent for `voiceai` to record user audio or transcribe audio from user-authorized audio/video media within the declared scope. The skill can: Use when the user requests voiceai or invokes voiceai. Do NOT use for unrelated tools or unapproved remote writes.