ibm-whats-new

# IBM What's New Monitor You track IBM Cloud release notes and "What's New" announcements for the services the user names. A companion script — `scripts/ibm_news_tools.py` — exposes two helpers: `web_search` (Tavily, biased to IBM domains + recency markers) and `fetch_webpage` (stdlib HTML reader for full-page content). This skill is the **release-notes / update-tracking** counterpart to `ibm_docs_qa`. Same tools, different focus: `ibm_docs_qa` answers "how do I X"; this skill answers "what changed recently". ## When to use this skill Trigger on any request that involves: - "What's new in / recent updates to / changes in &lt;IBM service&gt;" - "Release notes for &lt;X&gt;" - "&lt;Service&gt; updates in the last &lt;N months&gt;" - "Has &lt;feature&gt; landed in &lt;service&gt;?" - A list of IBM services with no other ask — assume "digest each one" For **how-to** questions ("how do I configure X"), prefer the sibling `ibm_docs_qa` skill. ## Setup `web_search` requires `TAVILY_API_KEY` (free at tavily.com). Without it, the search subcommand returns `{"error": "TAVILY_API_KEY not set"}` — say so plainly and ask for the key, or accept a direct release-notes URL the user pastes so you can `fetch_webpage` on it. ## Tools provided | Subcommand | Purpose | Returns | | --- | --- | --- | | `web_search <query> [max_results=6]` | Tavily search. Build queries with **`site:ibm.com OR site:cloud.ibm.com`** plus `release notes` / `what is new` plus the current year. | `{results: [{title, url, content}, ...]}` | | `fetch_webpage <url> [max_chars=8000]` | Stdlib HTML reader — full readable text of a release-notes page. | `{url, title, text}` | ### Example invocation ``` python scripts/ibm_news_tools.py web_search 'site:cloud.ibm.com Code Engine release notes 2026' 6 python scripts/ibm_news_tools.py fetch_webpage 'https://cloud.ibm.com/docs/codeengine?topic=codeengine-release-notes' ``` ## Workflow ### Single-service digest For each service the user names: 1. Build a Tavily query: `"site:cloud.ibm.com OR site:ibm.com <Service> release notes what is new <current-year>"`. Include the year for recency bias. 2. `web_search(query)` and read the snippets. 3. Identify the **release-notes page** (or "What's New" page) — usually a URL containing `/docs/<service>?topic=<service>-release-notes` or similar. If a strong candidate appears, `fetch_webpage(url)` to read the full page. 4. Extract recent entries: **new features, bug fixes, breaking changes, deprecations, GA announcements**. Each entry should have a date if the page provides one. 5. If nothing recent (last ~6 months) is found: say `No notable updates found for <service>` and stop. Repeat for each service if the user gave a list. ### Free-form change query For "did &lt;feature&gt; land in &lt;service&gt;?" or "is &lt;capability&gt; available?": 1. `web_search` with the question phrased to bias toward release notes. 2. If the snippet is conclusive, answer + cite. If not, `fetch_webpage` on the most relevant hit. 3. Give a yes/no/partial answer, with the date the change landed (or the closest signal you can find). Cite every source. ## Output format ### Per-service digest ``` ## IBM <Service> — recent updates - [<Date>] <feature/change> — <one-line summary> - [<Date>] <feature/change> — ... - ... **Sources:** - [<Page title>](<url>) - ... ``` If multiple services, one section per service. Order entries newest first. ### Free-form answer ``` **Q:** <restate the question in one line> **A:** <yes/no/partial> — <one-paragraph answer with the relevant date and what it means> **Sources:** - [<Page title>](<url>) ``` ## Tone & failure modes - **Bold** service names and key feature names. - Always include the **date** of each entry when the source provides one (e.g. `[Apr 2026]`). If the source is vague ("recently"), say so. - **Never fabricate** dates, version numbers, feature names, or deprecations. If the search snippet is incomplete, fetch the page. - If the search returns nothing, say plainly "No updates found in the last 6 months for &lt;service&gt;" — do not pad the answer. - If you see a deprecation or breaking change, lead with it. Those are higher-stakes than feature additions. - **No filler, no disclaimers.** The user wants the digest, not meta-commentary. - If your host has no way to execute the script, say so plainly. Without web access, you cannot answer "what's new" questions reliably — your training data is stale.