What does this skill do, and when should you use it?
Blog Discourse is the recency-and-engagement research skill within claude-blog. It uses platform-targeted WebSearch queries and freshness filters to examine recent public discussion across Reddit, X, YouTube, Hacker News, dev.to, Medium, and other platforms. It produces a structured DISCOURSE.md brief at the project root plus JSON that downstream writing workflows can consume. It is designed for trends, news analysis, product-update reactions, and topics where current practitioner or customer discussion matters, and complements rather than replaces the authority-focused blog-researcher.
It runs four keyword-trap checks before searching, then decomposes named-entity topics into primary, counter-perspective, practitioner, tangential-entity, and time-anchor queries. It uses WebSearch with platform site operators and 30-day or 90-day freshness windows, collecting each result's platform, URL, original title, snippet, date, and visible engagement proxy. It scans snippets for prompt-injection patterns and marks suspicious data. It then invokes scripts/discourse_research.py to parse the results JSON, cluster sources and themes, score recency and engagement, identify new themes and cross-platform consensus, and emit DISCOURSE.md plus structured JSON.
- A content team preparing a trend or news-analysis article needs current discussion from multiple public platforms.
- A marketer reviewing a product update wants recent customer and practitioner reactions to inform the next draft.
- A blogger writing a “state of X” article needs timely discussion signals alongside authority research.
- An editor wants to generate a discourse brief and feed it directly into blog brief, blog write, or blog strategy.
How do you install this skill?
- Core functionality depends on WebSearch and several overseas public platforms, with no evidence of reliable mainland-China reachability.
- DISCOURSE.md is written to the project root and may be loaded by downstream skills; untrusted content should still receive human review despite the stated injection defenses.
- Repository-level CI and test claims do not establish that this skill's key paths are verified; inspect scripts, dependencies, and output-file behavior before use.
- Shell / CLI
- Network access
- Local filesystem
Python 3.11+WebSearch
This skill is installed as part of the AgriciDaniel/claude-blog collection. With Claude Code 1.0.33+, run /plugin marketplace add AgriciDaniel/claude-blog, then /plugin install claude-blog@agricidaniel-claude-blog. The source does not document a separate installation procedure for skills/blog-discourse.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/blog-discourse" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
In Claude Code, run /blog discourse <topic> to research the last 30 days and create DISCOURSE.md at the project root. Use /blog discourse <topic> --days 90 for a wider window, or add --feed-into brief, --feed-into write, or --feed-into strategy to pass the brief into a downstream workflow. Use --input results.json when results have already been collected.
What are this skill's strengths and limitations?
- Requires no platform API keys and uses platform-targeted WebSearch instead.
- Covers several public discussion platforms while considering freshness and visible engagement.
- Includes keyword-trap checks, named-entity decomposition, prompt-injection marking, and schema validation.
- Produces both DISCOURSE.md and JSON and can feed brief, write, and strategy workflows.
- Clearly positioned as a complement to blog-researcher rather than a replacement for authority research.
- Depends on Claude Code `/blog` commands, orchestration, and WebSearch behavior, so other clients require adaptation.
- Search quality depends on public-page indexability, SERP snippets, and visible engagement data.
- When searches return no results, it reports insufficient coverage rather than inventing a brief.
- The source provides no independent test count or cross-platform validation evidence for this sub-skill.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Blog Discourse Research this page | 54 · Use with care | ★ 2.4k | 14d ago | MIT |
| Blog Strategy Architect | 49 · Use with care | ★ 2.4k | 14d ago | MIT |
| Blog Brief Planner | 44 · Not recommended | ★ 2.4k | 14d ago | MIT |
| Claude Blog Content Engine | 47 · Use with care | ★ 2.4k | 14d ago | MIT |
| SEO Content Audit Skill | 48 · Use with care | ★ 2.8k | 8mo ago | MIT |
Compared with blog-researcher, which focuses on authority sources and statistics, Blog Discourse focuses on recent practitioner discussion, engagement signals, and cross-platform consensus over a 30- or 90-day window. They are intended to be used together. It also differs from blog-notebooklm, which performs source-grounded research over user-uploaded documents.
How did FollowSkills review this skill?
The skill treats search snippets as untrusted data, requires prompt-injection pattern scanning, uses HTTPS-only and field validation, and specifies an unpredictable owner-only temporary file. It also discloses the WebSearch, project-root write, and downstream-loading data flow. Points are deducted because it relies on external search content and script execution without fully specifying user confirmation, cleanup, or rollback within this skill.
Commands, phases, parameters, zero-result behavior, and script-error handling are documented, including validation and temporary-file handling. Points are deducted because the skill provides no skill-specific test suite, compatibility matrix, or reproducible evidence for key paths; static review therefore cannot support a higher score.
The audience, trigger phrases, topic input, 30/90-day windows, feed-into options, and non-fit boundary for evergreen explainers are reasonably clear. Points are deducted because Chinese-language handling is unspecified and WebSearch reachability in mainland China is unverified; platform coverage and result quality may vary by region.
The documentation has clear progressive structure and includes commands, workflow, output shape, composition, error handling, license, and upstream attribution. Points are deducted because this skill has no explicit version, changelog, named maintainer, or independent update path, and some dependencies and path assumptions require repository context.
The skill defines structured DISCOURSE.md and JSON outputs covering new themes, consensus, niche themes, and actionable practitioner details, so it plausibly addresses recent-disourse research. Points are deducted because no skill-specific representative output or third-party execution evidence is supplied, and snippets, dates, and engagement fields may be incomplete and require manual review.
The source gives a detailed methodology, input schema, output schema, error branches, and security constraints, providing limited auditability; repository context also claims CI and tests. Points are deducted because discourse_research.py and discourse-specific tests are not included in the supplied evidence, so clustering, scoring, LAW compliance, and actual WebSearch behavior cannot be independently checked.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →