Release Notes

AI DIAL Admin Frontend — release-notes enhancer

The CI publishes a release for every tag with bullets that are just the PR titles. Those bullets carry a lot of dirt — conventional-commit prefixes (feat(scope):, fix(area):), repeated scope prefixes for related PRs on the same feature, misfiled items (feat(scope): landing under Other), and ## Other mixing consumer-relevant items (dep security bumps) with pure internal noise (skill scaffolding, changelog generation PRs). The stable releases at https://github.com/epam/ai-dial-admin-frontend/releases are what those raw notes look like after a human editorial pass. This skill reproduces that pass.

You are running in a forked, isolated context. Read and research freely — only the final summary you return reaches the main conversation. All file writes happen in this fork; the draft lands at claude/release-notes/<tag>-draft.md.

When to use

• "Enhance the release notes for 0.16.2-rc.0"
• "Look at the latest pre-release notes and refine them"
• "Help me adjust release notes for the current rc"
• "The CI just published <tag>, make it readable"

Do not trigger on requests like "what changed in 0.16.0?" — that is a recall question, not a notes-editing task.

Inputs

tag = $tag — the GitHub release tag to enhance (e.g. 0.16.1-rc.0, 0.17.0). If empty, pick the most recent tag from gh release list --limit 5 and confirm with the user before editing.

Workflow

1. Resolve target and reference styles

1. gh release view <tag> --json body,name,tagName — capture the raw CI notes.
2. gh release list --limit 10 — locate the previous tag of the same kind (last stable for a stable release, the predecessor rc for a delta rc.N+1).
3. gh release view <prev-stable-tag> --json body (and <prev-rc-tag> when relevant) — these are the style anchors. Match their terseness; one line per bullet.
4. git tag --list | sort -V + git log <prev-tag>..<tag> --oneline — full commit list for the range, to spot hotfix commits the CI dropped.

Section name normalization: The CI sometimes emits ## Feat instead of ## Features (e.g. patch releases). Always normalize to ## Features in the draft regardless of what CI wrote.

2. Pull source context for each bullet

For every bullet in the raw notes:

1. Parse out the trailing (Issue #<issue>) (#<PR>) or just (#<PR>). Canonical reference order is #<issue> (#<PR>).
2. gh pr view <PR> --json title,body,labels — read the PR body for the why and what-it-replaces; the title is too compressed.
3. For bullets without a PR number, find the commit with git log <prev-tag>..<tag> --oneline | grep -i <keywords> and git show <hash> — fold these into a related entry rather than leaving them standalone.

Grouping related PRs: When multiple PRs share the same issue number or are clearly follow-ups on the same feature (e.g. feat(deployments): MCP registry images source + feat(deployments): MCP registry image follow-ups + feat(deployments): pre-filter MCP registry servers), fold them into one bullet with all PR refs at the end: (#2927, #2986, #3034). Don't list every sub-PR as a separate line — large releases (40+ features) require aggressive grouping to stay readable.

3. Cross-check docs/INFRA-CHANGELOG.md and upgrade plan for deployment changes

The Deployment Changes section is built from primary sources:

• git diff <prev-tag>..<tag> -- docs/INFRA-CHANGELOG.md — env-var additions/removals/renames that landed in the range.
• Read docs/INFRA-CHANGELOG.md at the tag's version section for the canonical variable names and their descriptions.
• Check whether docs/upgrade-plans/<base-version>.md exists (e.g. for tag 0.17.0-rc.0 look for docs/upgrade-plans/0.17.0.md). If it does, link to it. If not, check docs/INFRA-CHANGELOG.md for what changed.
• For env-var canonical names, verify against the actual source code — PR bodies sometimes use pre-rename names, code wins.

4. Classify each bullet

The raw CI's ## Features / ## Fixes / ## Other partition is unreliable because it keys off the conventional-commit prefix. Reclassify by actual user impact:

Drop these entirely — no consumer-visible effect:

• Skill scaffolding and Claude Code tooling (chore: create ticket skill, chore: init claude documentation, chore: add release-notes skill).
• CI-generated changelog PRs (chore: add changelog for 0.16.0).
• Pure internal refactors, renames, and test-only changes.
• Merge remote-tracking branch … commits.
• CI-only changes (chore(ci): add release candidate branching) — unless a maintainer specifically needs to know, in which case keep under Other with a one-line rationale.
• Upgrade-plan generation PRs (chore: clear upgrade plan).

Keep in Other — items maintainers or operators care about:

• Security-adjacent dependency bumps (chore(deps): bump dompurify, chore(deps): bump follow-redirects).
• Infrastructure-relevant dependency upgrades.

If unsure whether to keep a bullet: would a customer or operator reading these notes care? If no, drop it.

5. Rewrite each kept bullet

Raw form: * feat(scope): description (Issue #NNN) (#NNN). Rewrite to:

* <Active-voice description of what changed> — <brief why-it-matters> #<issue> (#<PR>)

Rules in order of importance:

1. One line per bullet. No multi-paragraph descriptions.
2. Drop the conventional prefix (feat:, fix:, chore:, feat(scope):). Replace with prose.
3. Drop scope-as-noun repetition. Don't start the bullet with the scope name — feat(deployments): stop button → Stop button for deployments not Deployments — stop button.
4. Use a — em-dash for the "why" clause, not a hyphen or colon.
5. Backticks for code identifiers: env vars (NIM_ENABLED), config keys, component names, API paths.
6. Preserve issue + PR refs at the end in #<issue> (#<PR>) order. For grouped entries list all PRs: (#2927, #2986, #3034).
7. Prefix with [Preview] for preview-gated features (eval framework, MCP registry, etc. when still gated by an env var).
8. Flag regressions explicitly: (regression fix) for items restoring previously-working behavior.
9. Quote CVE IDs verbatim for security upgrades.
10. For features that graduate to GA, lead with Graduate <feature> to GA — <gate that's removed>.

Example transformations (this project's patterns)

# Grouping multiple related deployment PRs into one bullet:
- * feat(deployments): support images from MCP registry (Issue #2227) (#2927)
- * feat(deployments): MCP registry images source (Issue #2227) (#2986)
- * feat(deployments): MCP registry image follow-ups (Issue #2599) (#3034)
- * feat(deployments): pre-filter MCP registry servers (Issue #2227) (#2892)
- * feat(deployments): remove version from MCP registry source input (Issue #2227) (#2980)
- * feat(deployments): MCP registry grid details (Issue #2227) (#2987)
+ * [Preview] MCP Registry as deployment image source — browse and pin images from an MCP Registry server directly in the Deployments UI #2227 (#2892, #2927, #2980, #2986, #2987, #3034)

# Dropping prefix and em-dashing the why:
- * feat(list-view): add created_at and updated_at columns (Issue #3099) (#3104)
+ * `Created at` and `Updated at` columns in entity list views — sortable timestamps for all grid rows #3099 (#3104)

# Folding follow-up PRs on the same issue:
- * feat(applications): source field follow-up (Issue #2803) (#3058)
- * feat(applications): reuse SourceField (Issue #2803) (#3032)
- * feat(application): support container source (Issue #2803) (#3051)
+ * Container source support for Applications — configure a source container in the application form #2803 (#3032, #3051, #3058)

# Marking a regression fix:
- * fix(prompts): version switching broken (Issue #2858) (#2865)
+ * Restore prompt version switching (regression fix) #2858 (#2865)

# Dropping noise from Other:
- * chore: create ticket skill (#3033)                 ← drop entirely
- * chore: add changelog for 0.16.0 (#3075)            ← drop entirely
- * chore(ci): add release candidate branching (#2972)  ← drop entirely

# Keeping security dep bump in Other:
- * chore(deps): bump dompurify (#2973)
+ * Upgrade `dompurify` dependency (#2973)

# Preview tag based on env-var gating:
- * feat(eval): `Preview` release evaluation framework
+ * [Preview] Evaluation framework — run and review eval runs via `DIAL_EVAL_API_URL`; gated by env var

6. Build the Deployment Changes section

Add this section only when the range introduces env-var, behavioral, or schema changes that operators need to act on.

This project's format: If docs/upgrade-plans/<base-version>.md exists at the tag, the Deployment Changes section is a one- or two-sentence summary followed by a link to the upgrade plan file — not an inline table. This is the established pattern:

## Deployment Changes

This release includes high-priority changes. Please review the [full upgrade guide](https://github.com/epam/ai-dial-admin-frontend/blob/<tag>/docs/upgrade-plans/<base-version>.md) before proceeding.

Use the alert callout if the changes are critical:

## Deployment Changes

> [!IMPORTANT]
> This release requires operator action before upgrading. Please review the [full upgrade guide](https://github.com/epam/ai-dial-admin-frontend/blob/<tag>/docs/upgrade-plans/<base-version>.md).

How to populate it:

1. git diff <prev-tag>..<tag> -- docs/INFRA-CHANGELOG.md — what landed in the range.

2. Read the relevant version block in docs/INFRA-CHANGELOG.md for variable names, defaults, and descriptions.

3. If no docs/upgrade-plans/<base-version>.md exists yet, build the section inline instead of linking out: a table with columns Variable | Required | Default | Description, one row per env var that changed in the range. Pull Required from whether docs/INFRA-CHANGELOG.md marks the var **required** (or code shows no fallback), Default from the changelog's (default: ...) note, and Description from the changelog entry trimmed to one clause — don't repeat the [Preview] tag there, it belongs on the variable name.

   ## Deployment Changes
   
   | Variable | Required | Default | Description |
   |---|---|---|---|
   | `NIM_ENABLED` | No | `false` | Enables the NIM Model Serving option in Deployments |
   | `DIAL_CORE_API_URL` | Yes | — | DIAL Core API url |
   

4. Characterize severity: if all new env vars are optional and default-off, use plain text ("This release adds optional configuration flags") ahead of the table. If any is required or changes existing behavior, use > [!IMPORTANT] ahead of the table.

What does not belong here: Per-app feature configuration (things users configure in the UI, not operators in helm/env). Deployment Changes is strictly for env vars, behavioral shifts at the service level, and external system migrations.

7. Pre-release / delta handling

If the target is <X.Y.Z>-rc.N with N ≥ 1:

• Cover only what changed since the previous rc — do not consolidate predecessor notes.
• Drop sections with no entries in the delta.
• Do not prepend a "Delta since " pointer — the -rc.N suffix already signals it.

8. Save the draft (and optional editorial companion)

Write:

• claude/release-notes/<tag>-draft.md — the final notes, ready to paste into the GitHub release body. No preamble or commentary — just headings and bullets.
• claude/release-notes/<tag>-editorial-notes.md (optional, only when useful) — non-obvious calls worth surfacing:
  • Grouping decisions (which PRs were folded and why).
  • Items dropped, with one-line reason each.
  • Open questions (ambiguous categorization, missing upgrade plan, env-var name disagreement).

9. Verify nothing was pushed to GitHub

This skill never runs gh release edit, gh release create, or any write operation against the repo. Drafts only.

Output format

The file saved to claude/release-notes/<tag>-draft.md follows this shape exactly (including --- separators, which match the CI format):

## Features

* <one bullet per change or group>

---

## Fixes

* <one bullet per change>

---

## Other

* <only consumer- or maintainer-relevant items>

---

## Deployment Changes

This release includes high-priority changes. Please review the [full upgrade guide](https://github.com/epam/ai-dial-admin-frontend/blob/<tag>/docs/upgrade-plans/<base-version>.md) before proceeding.

Omit any section (including Deployment Changes) that has no entries. For a delta rc, omit empty sections rather than writing No changes. Section order: Features → Fixes → Other → Deployment Changes.

Return to the main conversation

Return a short summary — five lines or fewer:

• The draft path (claude/release-notes/<tag>-draft.md).
• Counts of bullets per section after enhancement and grouping.
• Groupings that happened (e.g. "folded 6 MCP registry PRs into 1 bullet").
• Reclassifications (e.g. "moved 2 from Other → Features").
• Items dropped (count, with one example).
• Whether Deployment Changes was added, and what it links to (or why it was omitted).
• Any open questions (missing upgrade plan, ambiguous item, env-var name conflict).

Example:

Drafted claude/release-notes/0.16.0-draft.md. 18 Features (grouped from 43 raw), 22 Fixes (grouped from 55 raw), 4 Other. Folded 6 MCP registry PRs into 1 bullet, 4 container-source PRs into 1. Reclassified 0 items. Dropped 4 internal items (skill scaffolding, CI changelog PR, RC branching chore). Added Deployment Changes linking to docs/upgrade-plans/0.16.0.md with > [!IMPORTANT] callout (NIM_ENABLED / HF_ENABLED are new required opt-in flags). Open: fix(deployments): multiple validation issues has no issue number — left as-is with a note in editorial file.

Safety rails

• Never edit GitHub. No gh release edit, no gh release create. Drafts only.
• Never invent items. Every kept bullet maps to a PR or a commit hash in the range.
• Never silently drop a PR reference. The bullet ends with the canonical #<issue> (#<PR>) refs.
• Verify canonical names from source, not from PR bodies.
• Don't consolidate pre-release notes into the stable's notes unless explicitly asked.
• Match the terseness of the predecessor's notes.

Maintenance

If you notice a pattern in the raw CI notes that this skill doesn't handle (a new CI section, a recurring rewrite the user keeps requesting, a new conventional-commit scope that misroutes items), surface it in your return summary and offer to update this SKILL.md. The user can confirm before any edit lands.