course

Name: course
Author: tuan3w

32stars2forks

How to install course?

tuan3w/obsidian-vault-agent/course

$npx -y skills add tuan3w/obsidian-vault-agent --skill course

Installs into the current project.

Summary

Process an online course into detailed vault notes — one per lecture, with slides and transcripts synthesized. Use this skill whenever the user shares a course URL, syllabus page, or lecture playlist and wants structured notes from it. Also triggers on "process this course", "tak

Files · 1

View on GitHub

SKILL.md

1	<Purpose>
2	Two modes:
3
4	Process mode (default): Crawl an online course page, extract the lecture
5	schedule with all materials (slides PDFs, YouTube videos, readings), then
6	process each lecture into a detailed vault-formatted note with embedded slide
7	images and synthesized transcripts. Creates a Course index note linking
8	everything together.
9
10	Refine mode: Re-read existing course lecture notes and improve them —
11	deepen equation explanations, add missing analogies, fix structure, strengthen
12	cross-references. Triggered when user says "refine", "improve", "fix", or
13	references existing course notes rather than a URL.
14
15	This is an orchestrator — it coordinates downloading, extraction, and synthesis
16	across multiple lectures, producing a complete course knowledge package in the
17	vault.
18	</Purpose>
19
20	<Use_When>
21	- User shares a course URL (syllabus, schedule, or homepage)
22	- User says "process this course" or "take notes on these lectures"
23	- User pastes a university course page with a lecture list
24	- User wants to batch-process a series of lectures from one course
25	- User has a YouTube playlist of course lectures
26	- User wants to refine/improve existing course notes ("refine mit diffusion notes", "improve L03")
27	</Use_When>
28
29	<Do_Not_Use_When>
30	- User has a single YouTube video (use /youtube)
31	- User has a single local video file (use /lecture)
32	- User wants to process an existing vault note that's NOT a lecture (use /process)
33	- User wants a single paper summarized (use /paper)
34	</Do_Not_Use_When>
35
36	<Steps>
37
38	## Stage 1: CRAWL — Extract the Lecture List
39
40	Parse the course URL from $ARGUMENTS. If no URL, ask the user.
41
42	Fetch the course page and extract structured lecture data:
43
44	```
45	WebFetch(
46	url="COURSE_URL",
47	prompt="Extract the complete course structure as JSON. For each lecture/session, include:
48	- number (int)
49	- title (string)
50	- date (string, if available)
51	- slides_url (string or null — look for PDF links to slides/lecture notes)
52	- video_url (string or null — look for YouTube links)
53	- readings (array of {title, url} — papers, blog posts, textbook chapters)
54	- description (string or null — any summary text)
55	Also extract:
56	- course_title (string)
57	- course_code (string or null)
58	- instructors (array of strings)
59	- course_url (string — the page URL)
60	- course_notes_url (string or null — if there's a single PDF of all course notes)
61	Return ONLY valid JSON, no markdown fencing."
62	)
63	```
64
65	Parse the JSON response. If the page has relative URLs for slides/videos, resolve
66	them against the course URL's base.
67
68	Handle edge cases:
69	- If the page is a YouTube playlist: extract video IDs and titles from the playlist
70	- If slides URLs are relative (e.g., `../docs/lecture_01.pdf`): resolve to absolute URLs
71	- If no structured schedule found: ask user to provide lecture list manually
72
73	## Stage 1b: DERIVE COURSE TAG
74
75	Generate a short, memorable course hashtag from the course code or title.
76	This tag will be used consistently across ALL notes for this course.
77
78	Rules for the tag:
79	- Use a descriptive short slug from the course topic, not just the code
80	- The tag should tell you what the course is ABOUT at a glance
81	- Examples: `#mit-diffusion`, `#cs231n-vision`, `#stanford-rl`, `#fast-ai-dl`
82	- Format: `institution-topic` or `code-topic` — lowercase, hyphens
83	- Keep it under 20 characters — short enough to type, long enough to understand
84	- Confirm the tag with the user before proceeding
85
86	Also derive a COURSE_SLUG for asset folder naming (same as tag without `#`).
87	Example: assets folder `assets/mit-diffusion/`, tag `#mit-diffusion`.
88
89	## Stage 2: PLAN — Confirm Scope with User
90
91	Present the extracted lecture list to the user in a clear table:
92
93	```
94	Found N lectures for "Course Title":
95	Course tag: #6s184
96	Assets folder: assets/6s184/
97
98	\| # \| Title \| Slides \| Video \| Readings \|
99	\|---\|-------\|--------\|-------\|----------\|
100	\| 1 \| Topic \| PDF \| YT \| 2 papers \|
101	\| 2 \| Topic \| PDF \| — \| 1 paper \|
102	...
103
104	Which lectures should I process? (default: all)
105	Options: "all", "1-3", "1,3,5", or specific numbers
106	```
107
108	If $ARGUMENTS includes a range (e.g., "1-3"), skip confirmation and use that range.
109
110	## Stage 3: PROCESS — Handle Each Lecture
111
112	### Content Source Priority
113
114	Not all sources are equal. A 50-minute video transcript wh

People who liked this also viewed

6 picks

Type

course

32stars2forks

View on GitHub

How to install course?

tuan3w/obsidian-vault-agent/course

$npx -y skills add tuan3w/obsidian-vault-agent --skill course

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	<Purpose>
2	Two modes:
3
4	Process mode (default): Crawl an online course page, extract the lecture
5	schedule with all materials (slides PDFs, YouTube videos, readings), then
6	process each lecture into a detailed vault-formatted note with embedded slide
7	images and synthesized transcripts. Creates a Course index note linking
8	everything together.
9
10	Refine mode: Re-read existing course lecture notes and improve them —
11	deepen equation explanations, add missing analogies, fix structure, strengthen
12	cross-references. Triggered when user says "refine", "improve", "fix", or
13	references existing course notes rather than a URL.
14
15	This is an orchestrator — it coordinates downloading, extraction, and synthesis
16	across multiple lectures, producing a complete course knowledge package in the
17	vault.
18	</Purpose>
19
20	<Use_When>
21	- User shares a course URL (syllabus, schedule, or homepage)
22	- User says "process this course" or "take notes on these lectures"
23	- User pastes a university course page with a lecture list
24	- User wants to batch-process a series of lectures from one course
25	- User has a YouTube playlist of course lectures
26	- User wants to refine/improve existing course notes ("refine mit diffusion notes", "improve L03")
27	</Use_When>
28
29	<Do_Not_Use_When>
30	- User has a single YouTube video (use /youtube)
31	- User has a single local video file (use /lecture)
32	- User wants to process an existing vault note that's NOT a lecture (use /process)
33	- User wants a single paper summarized (use /paper)
34	</Do_Not_Use_When>
35
36	<Steps>
37
38	## Stage 1: CRAWL — Extract the Lecture List
39
40	Parse the course URL from $ARGUMENTS. If no URL, ask the user.
41
42	Fetch the course page and extract structured lecture data:
43
44	```
45	WebFetch(
46	url="COURSE_URL",
47	prompt="Extract the complete course structure as JSON. For each lecture/session, include:
48	- number (int)
49	- title (string)
50	- date (string, if available)
51	- slides_url (string or null — look for PDF links to slides/lecture notes)
52	- video_url (string or null — look for YouTube links)
53	- readings (array of {title, url} — papers, blog posts, textbook chapters)
54	- description (string or null — any summary text)
55	Also extract:
56	- course_title (string)
57	- course_code (string or null)
58	- instructors (array of strings)
59	- course_url (string — the page URL)
60	- course_notes_url (string or null — if there's a single PDF of all course notes)
61	Return ONLY valid JSON, no markdown fencing."
62	)
63	```
64
65	Parse the JSON response. If the page has relative URLs for slides/videos, resolve
66	them against the course URL's base.
67
68	Handle edge cases:
69	- If the page is a YouTube playlist: extract video IDs and titles from the playlist
70	- If slides URLs are relative (e.g., `../docs/lecture_01.pdf`): resolve to absolute URLs
71	- If no structured schedule found: ask user to provide lecture list manually
72
73	## Stage 1b: DERIVE COURSE TAG
74
75	Generate a short, memorable course hashtag from the course code or title.
76	This tag will be used consistently across ALL notes for this course.
77
78	Rules for the tag:
79	- Use a descriptive short slug from the course topic, not just the code
80	- The tag should tell you what the course is ABOUT at a glance
81	- Examples: `#mit-diffusion`, `#cs231n-vision`, `#stanford-rl`, `#fast-ai-dl`
82	- Format: `institution-topic` or `code-topic` — lowercase, hyphens
83	- Keep it under 20 characters — short enough to type, long enough to understand
84	- Confirm the tag with the user before proceeding
85
86	Also derive a COURSE_SLUG for asset folder naming (same as tag without `#`).
87	Example: assets folder `assets/mit-diffusion/`, tag `#mit-diffusion`.
88
89	## Stage 2: PLAN — Confirm Scope with User
90
91	Present the extracted lecture list to the user in a clear table:
92
93	```
94	Found N lectures for "Course Title":
95	Course tag: #6s184
96	Assets folder: assets/6s184/
97
98	\| # \| Title \| Slides \| Video \| Readings \|
99	\|---\|-------\|--------\|-------\|----------\|
100	\| 1 \| Topic \| PDF \| YT \| 2 papers \|
101	\| 2 \| Topic \| PDF \| — \| 1 paper \|
102	...
103
104	Which lectures should I process? (default: all)
105	Options: "all", "1-3", "1,3,5", or specific numbers
106	```
107
108	If $ARGUMENTS includes a range (e.g., "1-3"), skip confirmation and use that range.
109
110	## Stage 3: PROCESS — Handle Each Lecture
111
112	### Content Source Priority
113
114	Not all sources are equal. A 50-minute video transcript wh

People who liked this also viewed

6 picks

Type