blogwatcher

Name: blogwatcher
Author: AlexAI-MCP

132stars24forks

How to install blogwatcher?

alexai-mcp/hermes-ccc/blogwatcher

$npx -y skills add AlexAI-MCP/hermes-CCC --skill blogwatcher

Installs into the current project.

Summary

Monitor and summarize blog posts, RSS feeds, and web content for research and staying current with topics.

Files · 1

View on GitHub

SKILL.md

1	# Blogwatcher
2
3	## Purpose
4
5	- Use this skill to monitor RSS feeds and blog-style content for ongoing research.
6	- It is useful for staying current on AI, ML, engineering, and product ecosystems.
7	- Prefer RSS when you want structured updates without scraping full websites every time.
8
9	## Install
10
11	```bash
12	pip install feedparser
13	```
14
15	## Parse a Feed
16
17	```python
18	import feedparser
19
20	feed = feedparser.parse("https://example.com/rss")
21	print(feed.feed.title)
22	print(feed.entries[0].title)
23	```
24
25	- `feedparser` handles RSS and Atom feeds.
26	- The parsed object exposes feed metadata and a list of entries.
27
28	## Access Entry Fields
29
30	```python
31	entry = feed.entries[0]
32	print(entry.title)
33	print(entry.summary)
34	print(entry.link)
35	print(entry.published)
36	```
37
38	- The most commonly useful fields are `.title`, `.summary`, `.link`, and `.published`.
39	- Not every feed includes every field, so code defensively.
40
41	## Batch Fetch Multiple Feeds
42
43	```python
44	import feedparser
45
46	feeds = [
47	"https://example.com/rss",
48	"https://another.example/feed.xml",
49	]
50
51	all_entries = []
52	for url in feeds:
53	parsed = feedparser.parse(url)
54	for entry in parsed.entries:
55	all_entries.append(
56	{
57	"source": parsed.feed.get("title", url),
58	"title": entry.get("title", ""),
59	"summary": entry.get("summary", ""),
60	"link": entry.get("link", ""),
61	"published": entry.get("published", ""),
62	}
63	)
64	```
65
66	- Batch fetches are the standard pattern for topic monitoring.
67	- Normalize fields early so downstream summarization stays simple.
68
69	## Filter by Date or Keyword
70
71	- Filter by date when you only care about the most recent week or month.
72	- Filter by keyword when watching narrow topics like `agents`, `evals`, or `multimodal`.
73	- Keep the filter stage simple and deterministic.
74
75	Example:
76
77	```python
78	keywords = ["llm", "agents", "retrieval"]
79	filtered = [
80	e for e in all_entries
81	if any(k.lower() in (e["title"] + " " + e["summary"]).lower() for k in keywords)
82	]
83	```
84
85	## Summarize With Claude
86
87	- Extract the title and summary from each entry.
88	- Feed the collected set into Claude and ask for synthesis by theme, signal, and novelty.
89	- This works better than summarizing one feed item at a time when you are tracking a field.
90
91	Example prompt shape:
92
93	```text
94	Summarize these AI research updates. Group them into model releases, tooling, benchmarks, and policy. Highlight what appears genuinely new.
95	```
96
97	- Keep the raw title, summary, link, and publication date in the source material.
98	- Ask for a synthesis, not a rewrite.
99
100	## Save Results to a File
101
102	```python
103	import json
104
105	with open("feed_digest.json", "w", encoding="utf-8") as f:
106	json.dump(filtered, f, ensure_ascii=False, indent=2)
107	```
108
109	- Save structured digests for later comparison.
110	- JSON is the easiest format for later reprocessing.
111	- Markdown is convenient if the output is intended for direct reading.
112
113	## Common AI and ML Blogs To Monitor
114
115	- arXiv Sanity
116	- Hugging Face Blog
117	- OpenAI
118	- Anthropic
119	- engineering blogs from inference providers, vector DB vendors, and cloud platforms
120
121	- Some sources are better via RSS.
122	- Others may require periodic scraping or newsletter ingestion.
123
124	## Combine With `/arxiv`
125
126	- Use this skill for blog and announcement monitoring.
127	- Combine it with `/arxiv` for paper discovery and academic monitoring.
128	- The combination gives better coverage across research papers, product launches, and engineering writeups.
129
130	## Good Monitoring Workflow
131
132	1. Define a small set of feeds by topic.
133	2. Fetch them on a schedule.
134	3. Normalize fields into one list.
135	4. Filter by keyword and freshness.
136	5. Ask Claude to synthesize the daily or weekly signal.
137	6. Save the resulting digest for later reference.
138
139	## Practical Notes
140
141	- RSS coverage varies widely by site.
142	- Feed summaries are often enough for triage but not enough for deep analysis.
143	- Follow links for the few items that survive filtering.
144	- Save source links alongside the synthesis so claims remain traceable.
145
146	## Summary
147
148	- Install RSS parsing support with `pip install feedparser`.
149	- Parse feeds with `feedparser.parse("https://example.com/rss")`.
150	- Access fields like `feed.entries[0].title`, `.summary`, `.link`, and `.published`.
151	- Batch fetch multiple feeds, then filter by date or keyword.
152	- Summarize title and summary pairs with Claude for cross-source synthesis.
153	- Save digests to a file and combine this workflow with `/arxiv` for broader research monitoring.

People who liked this also viewed

6 picks

Type

blogwatcher

132stars24forks

View on GitHub

How to install blogwatcher?

alexai-mcp/hermes-ccc/blogwatcher

$npx -y skills add AlexAI-MCP/hermes-CCC --skill blogwatcher

Installs into the current project.

Summary

Monitor and summarize blog posts, RSS feeds, and web content for research and staying current with topics.

Files · 1

View on GitHub

SKILL.md

1	# Blogwatcher
2
3	## Purpose
4
5	- Use this skill to monitor RSS feeds and blog-style content for ongoing research.
6	- It is useful for staying current on AI, ML, engineering, and product ecosystems.
7	- Prefer RSS when you want structured updates without scraping full websites every time.
8
9	## Install
10
11	```bash
12	pip install feedparser
13	```
14
15	## Parse a Feed
16
17	```python
18	import feedparser
19
20	feed = feedparser.parse("https://example.com/rss")
21	print(feed.feed.title)
22	print(feed.entries[0].title)
23	```
24
25	- `feedparser` handles RSS and Atom feeds.
26	- The parsed object exposes feed metadata and a list of entries.
27
28	## Access Entry Fields
29
30	```python
31	entry = feed.entries[0]
32	print(entry.title)
33	print(entry.summary)
34	print(entry.link)
35	print(entry.published)
36	```
37
38	- The most commonly useful fields are `.title`, `.summary`, `.link`, and `.published`.
39	- Not every feed includes every field, so code defensively.
40
41	## Batch Fetch Multiple Feeds
42
43	```python
44	import feedparser
45
46	feeds = [
47	"https://example.com/rss",
48	"https://another.example/feed.xml",
49	]
50
51	all_entries = []
52	for url in feeds:
53	parsed = feedparser.parse(url)
54	for entry in parsed.entries:
55	all_entries.append(
56	{
57	"source": parsed.feed.get("title", url),
58	"title": entry.get("title", ""),
59	"summary": entry.get("summary", ""),
60	"link": entry.get("link", ""),
61	"published": entry.get("published", ""),
62	}
63	)
64	```
65
66	- Batch fetches are the standard pattern for topic monitoring.
67	- Normalize fields early so downstream summarization stays simple.
68
69	## Filter by Date or Keyword
70
71	- Filter by date when you only care about the most recent week or month.
72	- Filter by keyword when watching narrow topics like `agents`, `evals`, or `multimodal`.
73	- Keep the filter stage simple and deterministic.
74
75	Example:
76
77	```python
78	keywords = ["llm", "agents", "retrieval"]
79	filtered = [
80	e for e in all_entries
81	if any(k.lower() in (e["title"] + " " + e["summary"]).lower() for k in keywords)
82	]
83	```
84
85	## Summarize With Claude
86
87	- Extract the title and summary from each entry.
88	- Feed the collected set into Claude and ask for synthesis by theme, signal, and novelty.
89	- This works better than summarizing one feed item at a time when you are tracking a field.
90
91	Example prompt shape:
92
93	```text
94	Summarize these AI research updates. Group them into model releases, tooling, benchmarks, and policy. Highlight what appears genuinely new.
95	```
96
97	- Keep the raw title, summary, link, and publication date in the source material.
98	- Ask for a synthesis, not a rewrite.
99
100	## Save Results to a File
101
102	```python
103	import json
104
105	with open("feed_digest.json", "w", encoding="utf-8") as f:
106	json.dump(filtered, f, ensure_ascii=False, indent=2)
107	```
108
109	- Save structured digests for later comparison.
110	- JSON is the easiest format for later reprocessing.
111	- Markdown is convenient if the output is intended for direct reading.
112
113	## Common AI and ML Blogs To Monitor
114
115	- arXiv Sanity
116	- Hugging Face Blog
117	- OpenAI
118	- Anthropic
119	- engineering blogs from inference providers, vector DB vendors, and cloud platforms
120
121	- Some sources are better via RSS.
122	- Others may require periodic scraping or newsletter ingestion.
123
124	## Combine With `/arxiv`
125
126	- Use this skill for blog and announcement monitoring.
127	- Combine it with `/arxiv` for paper discovery and academic monitoring.
128	- The combination gives better coverage across research papers, product launches, and engineering writeups.
129
130	## Good Monitoring Workflow
131
132	1. Define a small set of feeds by topic.
133	2. Fetch them on a schedule.
134	3. Normalize fields into one list.
135	4. Filter by keyword and freshness.
136	5. Ask Claude to synthesize the daily or weekly signal.
137	6. Save the resulting digest for later reference.
138
139	## Practical Notes
140
141	- RSS coverage varies widely by site.
142	- Feed summaries are often enough for triage but not enough for deep analysis.
143	- Follow links for the few items that survive filtering.
144	- Save source links alongside the synthesis so claims remain traceable.
145
146	## Summary
147
148	- Install RSS parsing support with `pip install feedparser`.
149	- Parse feeds with `feedparser.parse("https://example.com/rss")`.
150	- Access fields like `feed.entries[0].title`, `.summary`, `.link`, and `.published`.
151	- Batch fetch multiple feeds, then filter by date or keyword.
152	- Summarize title and summary pairs with Claude for cross-source synthesis.
153	- Save digests to a file and combine this workflow with `/arxiv` for broader research monitoring.

People who liked this also viewed

6 picks

Type