This closing lesson adds one more stage to the pattern from Lessons 4.3 and 4.6: gather, compile, and now produce — turning a pile of articles and news into one finished, on-brand piece of content, ready to publish. It is the same shape you've now built three times, extended one step further.
Content generation usually starts with research: reading a dozen articles to find the two or three worth writing about. Automating that research step — and handing the writer a compiled, organized digest instead of a dozen open tabs — is where this workflow earns its time back.
1Dictate the Workflow
Describe the whole pipeline in one message — sources, target content, and tone — the same dictate-to-setup pattern used throughout this module.
- Create a new folder, for example content-pipeline, and open it in Claude Code.
- Adapt and send the example dictation below.
- Answer any clarifying questions.
- Review the proposed CLAUDE.md before approving.
I want a workflow that turns news articles into one piece of content I can publish. Here are my sources: [list your 3-6 news sites, blogs, or feeds]. I care about articles touching these themes: [list your themes, e.g., industry trends, regulatory changes, competitor moves]. Once new articles come in, I want them grouped into one organized digest by theme, not just a flat list. Then I want a draft piece of content built from that digest — specifically a [blog post / LinkedIn post / newsletter section], around [word count], in our brand voice: [describe your tone, e.g., confident, plain-spoken, no hype]. Set this up as a project with a README and a CLAUDE.md capturing my sources, themes, content format, and tone, and ask me anything you need first.
What Claude Code should produce:
content-pipeline/
├── README.md Explains the pipeline
├── CLAUDE.md Standing rules: sources, themes,
│ content format, brand voice
├── data/
│ └── seen-articles-log.md Every article already gathered
├── digests/
│ └── theme-digests/ Compiled, theme-grouped digests
└── output/
└── drafts/ Finished content drafts land here
If you already built a Marketing Project in Lesson 3.7, its custom instructions — brand voice, banned words, tone — describe almost exactly what this CLAUDE.md needs. Copy from it instead of starting blank.
2Build the Three-Skill Plugin
Three skills, the same shape as Lesson 4.3's gather-respond-assemble pattern: gather-articles, compile-digest, draft-content. This reuses the plugin file structure from Lesson 4.3, so this step focuses on the three skills themselves.
- Ask Claude Code to build gather-articles, review it.
- Ask for compile-digest, review it.
- Ask for draft-content, review it.
- Run all three commands in order against real sources.
Skill 1 — /gather-articles. What to tell Claude Code to build:
Build a skill called gather-articles. Its job: fetch each source listed in CLAUDE.md, extract every article's title, date, and a short summary, and compare against data/seen-articles-log.md. Keep only articles not already logged and matching one of our themes in CLAUDE.md. Append each new, on-theme article to the log with its title, date, theme, source link, and summary.
What this produces (skills/gather-articles/SKILL.md, abbreviated):
--- name: gather-articles description: Fetch each source, extract new on-theme articles, and log them. Use when the user says "gather articles" or "/gather-articles." --- # gather-articles 1. Read the source list and theme list from CLAUDE.md. 2. Fetch each source with WebFetch and extract each article's title, date, and a short summary. 3. Read data/seen-articles-log.md — everything logged so far. 4. Keep only articles not already logged and matching a theme. 5. Append each one to the log with title, date, theme, source link, and summary. 6. Report how many new, on-theme articles were found.
Skill 2 — /compile-digest. What to tell Claude Code to build:
Build a skill called compile-digest. Its job: read every article in data/seen-articles-log.md not yet compiled, group them by theme, and write one organized digest file to digests/theme-digests/ with a section per theme, each article listed with its summary and source link. Mark each included article "compiled" in the log.
What this produces (skills/compile-digest/SKILL.md, abbreviated):
--- name: compile-digest description: Group uncompiled logged articles by theme into one organized digest file. Use when the user says "compile the digest" or "/compile-digest." --- # compile-digest 1. Read data/seen-articles-log.md. Find every article marked "gathered" but not yet "compiled." 2. Group them by theme. 3. Write digests/theme-digests/<date>-digest.md: one section per theme, each article's title, summary, and source link. 4. Mark each included article "compiled" in the log. 5. Report the digest's file path and how many themes it covers.
Skill 3 — /draft-content. What to tell Claude Code to build:
Build a skill called draft-content. Its job: read the latest digest in digests/theme-digests/, and draft one piece of content in the format and tone described in CLAUDE.md — a blog post, LinkedIn post, or newsletter section, at the target length also in CLAUDE.md. Reference specific points from the digest rather than writing generically. If the target format is a Word document, use the docx skill to save it to output/drafts/; otherwise save it as Markdown.
What this produces (skills/draft-content/SKILL.md, abbreviated):
--- name: draft-content description: Draft one piece of content from the latest theme digest, in the format and voice set in CLAUDE.md. Use when the user says "draft the content" or "/draft-content." --- # draft-content 1. Read the most recent file in digests/theme-digests/. 2. Read the target format, length, and brand voice from CLAUDE.md. 3. Draft the content, referencing specific articles and themes from the digest rather than generic statements. 4. If the target format is a Word document, use the docx skill to save output/drafts/<date>-draft.docx. Otherwise save output/drafts/<date>-draft.md. 5. Show the draft and note which digest it was built from.
Sample output:
/gather-articles → "4 new on-theme articles logged across 2 themes."
/compile-digest → "Compiled digests/theme-digests/tuesday-digest.md
— 2 themes: Industry Trends (3 articles),
Regulatory Changes (1 article)."
/draft-content → "Drafted output/drafts/tuesday-draft.md — a
420-word blog post referencing all 3 Industry
Trends articles and the regulatory update.
Saved in brand voice per CLAUDE.md."
The draft didn't start from a blank page — it started from an already-organized, already-themed digest. That's the entire value of splitting gather, compile, and draft into three separate skills instead of one.
3Install and Run It
Same install pattern as every plugin in this module.
- Register the local marketplace and install the plugin.
- Restart Claude Code.
- Run the three commands in order.
/plugin marketplace add /path/to/content-pipeline /plugin install content-pipeline@content-pipeline
Then, after restarting Claude Code inside the project folder:
/gather-articles /compile-digest /draft-content
draft-content produces a first draft, not a finished piece. Read it against the source articles, fact-check anything specific, and edit the tone before it goes anywhere public — the same validation habit from Lesson 1.3 applies here.
4Optional — Run It on a Schedule
The gather-and-compile steps can run unattended; drafting is worth doing with a person still reviewing the result.
Windows
- Before proceeding, revisit Lesson 4.1, Step 4 — confirm you trust both scheduled skills' output from manual runs first.
- Open Windows Task Scheduler.
- Schedule the two gathering commands to run daily or weekly.
- Leave /draft-content for you to run by hand, once you've reviewed the digest.
claude -p "/gather-articles" --cwd "C:\path\to\content-pipeline" claude -p "/compile-digest" --cwd "C:\path\to\content-pipeline"
Mac
- Before proceeding, revisit Lesson 4.1, Step 4 — confirm you trust both scheduled skills' output from manual runs first.
- Open Terminal and run crontab -e (opens in the nano editor by default).
- Add both lines below on a daily or weekly cadence, then save (Ctrl+O, Enter, Ctrl+X in nano). Leave /draft-content for you to run by hand, once you've reviewed the digest.
0 8 * * * claude -p "/gather-articles" --cwd "$HOME/content-pipeline" 5 8 * * * claude -p "/compile-digest" --cwd "$HOME/content-pipeline"
cron jobs on macOS often fail silently unless the terminal app running them has Full Disk Access. Go to System Settings → Privacy & Security → Full Disk Access and enable it for Terminal (or your terminal app).
Running all three commands yourself already works. Automating only the first two, and keeping the draft step manual, is a deliberate choice — not a requirement.
Automating research is low-risk — worst case, an irrelevant article gets logged. Automating what gets published carries more risk, so this lesson keeps a person deciding when to draft and reviewing before anything goes out.
On Windows: disable or delete the two scheduled tasks in Windows Task Scheduler. On Mac: run crontab -e and delete or comment out both lines. Either way, also disable the plugin the same way shown in Lesson 4.3, Step 6, if you want to stop entirely. Your logged articles and past digests stay exactly as they are either way.
Verification
If the draft reads generically instead of referencing real articles, confirm draft-content is actually reading the digest file, not writing from general knowledge.
Related Resources
- claude.ai — The Claude AI platform
- code.claude.com/docs — Official documentation
- support.claude.com — Help center and account support
Course complete
You have built four real automations across four different data sources — a webpage, email and calendar, public feeds, and news articles — all using the same gather-respond-assemble shape. Gather, compile, produce. Check, draft, assemble. Scan, log, digest. That repeating pattern is what to carry into your own next workflow, more than any single command. Pick one recurring task from your own role this course hasn't covered yet, and build a plugin around it — and whatever you build, make sure you know how to turn it off.