Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add AZIO-126/azio-vault-os --skill research-pipeline-v2git clone --depth 1 https://github.com/AZIO-126/azio-vault-osWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/azio-126/azio-vault-os/research-pipeline-v2)<a href="https://agentmods.dev/skills/azio-126/azio-vault-os/research-pipeline-v2"><img src="https://agentmods.dev/badge/skills/azio-126/azio-vault-os/research-pipeline-v2/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/azio-126/azio-vault-os/research-pipeline-v2"><img src="https://agentmods.dev/badge/skills/azio-126/azio-vault-os/research-pipeline-v2.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00160 | $0.04827 |
| Opus 5 | $0.00080 | $0.02413 |
| Sonnet 5 | $0.00032 | $0.00965 |
| Haiku 4.5 | $0.00016 | $0.00483 |
Grade A, and why
research-pipeline-v2 scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 467 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Research Pipeline v2 — Role-Based Curriculum Builder
Five roles write your curriculum in phases: ASR discovers what exists, NLM organizes it, Writer fills it, Deep-diver thickens it, Reviewer validates the learning journey.
Phase 1: Resource Collection → YouTube ASR/papers/知乎 (parallel)
Phase 2: 主题发现者 → ASR全文提取子主题+关键实体 (NEW)
Phase 3: 框架设计师 → 子主题+NLM → 框架+原子标题+多模态标记 (NEW)
↳ 用户确认框架
Phase 4: 教案撰稿人 → 框架+ASR全文 → 详细填充 (parallel agents)
Phase 5: 技术深挖专家 → enrich thin sections with examples/code
Phase 6: 学习动线审核员 → 框架动线审核 + 内容密度审核 + 直接修复 (ENHANCED)
Phase 7: Integration → cross-doc consistency, Overview, wikilinks
Why this order matters (A/B test finding)
- NLM-first (A): Good structure but shallow content — NLM generalizes away the specifics
- ASR-first (B): Dense content but no learning path — jumps into details without context
- This pipeline (C): ASR discovers topics → NLM designs the learning path → ASR fills the details → Best of both
Tool Paths
yt-dlp: yt-dlp
notebooklm: notebooklm
vault: {{VAULT_PATH}}
paper_search: {{SCRIPTS_PATH}}/paper_search.py
paper_score: {{SCRIPTS_PATH}}/paper_score.py
Phase 0: Extract Intent + Vault Scan
Before anything else:
Step 1: Vault scan
Grep vault for existing knowledge (concepts/, research/, projects/, sources/). Report findings.
Step 2: Extract intent
From user message: Goal, Deadline, Prior knowledge (vault + ask), Depth needed.
Phase 1: Resource Collection (parallel)
Launch ALL resource collection BEFORE designing the framework. The framework must be grounded in what the sources actually contain, not what we assume they contain.
2a. YouTube Search + ASR Download
# Search
yt-dlp --dump-json --flat-playlist "ytsearch15:[TOPIC]" 2>/dev/null | python3 -c "..."
# ASR download — MUST use --cookies-from-browser chrome
yt-dlp --cookies-from-browser chrome \
--write-auto-sub --sub-lang en,zh-Hans,zh --sub-format srt \
--skip-download --output "/tmp/transcripts/[subtopic]/%(title)s.%(ext)s" "URL"
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 467 lines · 160 tokens per session scan A 5be9ac7ea000
research-pipeline-v2 is a skill published in the GitHub repository AZIO-126/azio-vault-os (2 stars, last pushed 5mo ago), licensed MIT. It adds 160 tokens to every session and 4,827 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
VaultForge
Process learning materials into structured Obsidian atomic notes with roadmaps, MOCs, and double-links. Auto-generate study vaults from PDFs, Markdown, Word, PowerPoint, and more with AI-powered note-taking, knowledge linking, and deep research. Supports English and Chinese output. Use when the user wants to build a…
cognitive-compile
Use when source material must be transformed into a compact, evidence-aware model for learning, decisions, or an Obsidian concept note.
notebooklm
Drive NotebookLM over a programmatic CLI with no browser: turn a vault slice into an audio overview, study guide or mind map, and fold the artifacts back into the vault. Browser automation is the fallback if CLI auth breaks. Triggers: "/notebooklm", "make an audio digest of ", "podcast from the vault".
Research for Datacore
Automated research — source processing, podcast generation, and knowledge extraction.
wiki-ingest
Ingest supplied source material into an Obsidian vault with provenance and claim tracking: pasted text, files staged in the selected vault's inbox or .raw archive, or explicitly approved URLs. Use for a single source or bounded batch, not for saving an assistant answer. Triggers: ingest, ingest this file, ingest this…
autoresearch
Run a bounded, source-grounded research loop, draft a cited dossier, and optionally propose a separately reviewed canonical vault merge. Use when the user wants autonomous or deep research that may access the public web. Triggers: /autoresearch, autoresearch, research this topic, deep dive into, investigate, find…