Day 25 opened on Saturday April 4th. Clean restart with tar package including
2DO-LIST-DAY25.html, 2026-03-31-PROJECT-DIARY.html,
about.html, ripgrep.html, deep-dive-index.html,
and cli-tools-start-here.html.
Session opened with a checker run showing one orphan —
collaboration/addendum/2026-04-04_Day25_Notes.txt — which is
intentional. The addendum directory and file are already referenced in the Day 25
diary card. The orphan resolves when the card is inserted into
project-diary.html. No corrective action required.
Primary content work was the curl deep-dive page. curl was the designated Day 25
primary deliverable per the 2DO list. The AI integration angle is strong: every major
AI API — Anthropic, OpenAI, local ollama — is reachable with curl from
the shell. No SDK, no Python environment, no additional dependencies. The page covers
direct Anthropic API calls, a reusable ask_claude.sh script, the ollama
REST API, the full rg→jq→curl→jq pipeline for air-gapped log analysis,
and a shell function wrapper for quick AI access from any prompt.
The LearnLinux candidates (bat, jc, sed/awk primer) were identified but deferred to Day 26 — curl was the correct priority and the session closed cleanly without overreaching.
Full CLI tool deep-dive page for curl. Sections: What Is curl, Installation (dnf/apt, version note for AlmaLinux 9 7.76+), Basic Usage (17-row flag reference table, GET/POST/auth patterns), SysAdmin Use Cases (health checking with timing breakdown, TLS certificate checking, downloads with integrity verification, internal API testing, monitoring scripts, Kubernetes API), curl + AI Tools (Anthropic API directly, ask_claude.sh script, ollama REST API, full rg+jq+curl+jq pipeline, OpenAI-compatible endpoints, curl+jq pairing, curl+rg fetch-and-filter, reusable ask() shell function), ~/.curlrc configuration, Honest Assessment, Related Tools. 802 lines. vnu clean.
curl card added to the 🐧 AI for Unix/Linux section, after the ripgrep card. Last updated date bumped to April 4, 2026. vnu clean.
Four additions to integrate curl into the toolchain documentation:
Last updated date bumped to April 4, 2026. vnu clean.
LearnLinux candidates correctly deferred from Day 25 — ready for Day 26:
Format templates needed from LearnLinux before generating: one CMDS page and one EXPANDED page to match established style.
curl.html, deep-dive-index.html,
cli-tools-start-here.html — all clean, three for threefind . -name "*.html~" -print | xargs rm — cleancheck_ai_tree.sh -a — 1 orphan (addendum, intentional —
resolves on diary card insertion), 0 broken linksThe curl page landed cleanly on the first vnu pass — no errors. The ripgrep format is now well-established enough that a new tool page can be generated and validated in a single session without rework cycles.
The AI integration angle for curl is genuinely strong. The mental model shift is
simple: AI APIs are REST APIs. SysAdmins already know how to call REST APIs with curl.
The only new elements are the API key in the header and the JSON body structure. The
jq -Rs . trick for converting multi-line stdin to a JSON string is the
one technique that needs explicit teaching — the page covers it.
The addendum directory pattern established today is useful. Session notes in
collaboration/addendum/ linked from the diary card keep working context
inside the site structure without polluting the main content directories.
cmh — AI Education Project Day 25 — April 4, 2026
Next session: Day 26 — LearnLinux bat/jc/sed-awk candidates