blog-discourse
Installation
SKILL.md
Blog Discourse: Real Discourse Research, API-Free
Produces DISCOURSE.md: a structured brief of what practitioners said about on the public web in the last 30 days. It is the recency + engagement lens that blog-researcher (authority-first) lacks, asking what practitioners and customers are actually saying about this topic right now.
Adapted from the methodology of last30days-skill (Matt Van Horn, MIT, https://github.com/mvanhorn/last30days-skill). The upstream uses platform APIs; this sub-skill uses WebSearch with platform-targeted site operators. No API keys required.
Commands
| Command | Purpose |
|---|---|
/blog discourse <topic> |
Produce a discourse brief at project-root DISCOURSE.md |
/blog discourse <topic> --days 90 |
Widen the freshness window from 30 to 90 days |
/blog discourse <topic> --input results.json |
Skip search; build the brief from a pre-gathered results file. The flag name matches scripts/discourse_research.py --input directly. |
/blog discourse <topic> --output path.md |
Write markdown to a chosen output path and print structured JSON without markdown to stdout. |
/blog discourse <topic> --format json |
Print the full JSON brief to stdout when no --output path is used. |
/blog discourse <topic> --decomposition questions.txt |
Pass newline-delimited decomposition questions into the helper. |