[fsaresh]

Why

While a question dialog is open, the user may see only the question: its header, text and option labels. The message above it can be hidden. Two kinds of question in these skills asked about content that appeared only in the message above:

• "Run this command" in the release plans skill, where the user could not see the curl command.
• "Looks good" in onboarding Step 7, which approves a plan the user could not see.

#267 adds the general rule to CONTRIBUTING.md and the skill template.

What changed

launchdarkly-release-plans (1.0.0 → 1.0.1)

• "User-run commands": the command also goes in the question. It goes in the preview of "Done, here's the output" when the question tool has previews (AskUserQuestion), or else in the question text. It also stays in the message.
• New core principle 2: "Use this plan", "Looks right", "Make this change", "Update " and "Run this command" carry the plan, change or command. This holds in every mode, including Record only.
• The starter plan "Looks right" steps, in SKILL.md and references/environment-release-plans.md, point at that principle.

launchdarkly-release-autopilot-onboarding (1.1.0 → 1.1.1)

• The "Links" rule is now "The question carries what it asks about":
  • Links go in the question text, as before.
  • A plan, summary, command or diff goes in the preview of the option that accepts it. Without previews, it goes in the question text.
• Step 1: the summary of the current setup goes in the "Add more / Change one setting / Diagnose a PR" question.
• Step 6 and Step 8: the "Run this command" steps follow the release plans skill's rule.
• Step 7: the full plan goes in the preview of "Looks good". The skill says never to ask with only "Ready to apply?".
• Step 8: the second "Looks good" after a changed flag release plan carries the changed plan.
• Step 9: the test PR asks header "Open PR" with the diff in the preview of "Open it".

Evals

• New runQuestionShowsCommand check, added to the six cases that show REST commands (four in release plans, two in onboarding). Every "Done, here's the output" question must show a curl command in its text or in an option's preview.
• New confirmShowsPlan check, added next to confirmsBeforeWriting in the 15 onboarding cases that use it. Every "Looks good" question must show the plan as a list, and the plan must name every repository that the run later writes.
• New unit tests for both checks.

Testing

• node --test tests/*.test.js:
  • 146 tests pass, including the new tests.
  • 7 tests fail the same way on main, because zod and js-yaml are not installed in my checkout.
• scripts/validate_skills.py, python3 -m unittest discover -s tests and generate_catalog.py --check pass. skills.json is regenerated for the version bumps.
• Both promptfooconfig.yaml files parse.
• I did not run the promptfoo eval suites. The new checks add requirements to 20 eval cases, so run both suites before merge.

Note

#264 and #265 change the same eval files, so whichever merges later needs a small rebase.

🤖 Generated with Claude Code

_{Opened from termag}

---

Note

Overview
Release plans and onboarding skills now require that question dialogs include the content they ask about (plans, curl commands, diffs, links)—via option preview or question text—because users may not see the message above while a question is open.

launchdarkly-release-plans (1.0.0 → 1.0.1) adds a core principle and expands User-run commands so "Run this command" / "Done, here's the output" always shows the curl block in the question; starter "Looks right" steps reference the same rule.

launchdarkly-release-autopilot-onboarding (1.1.0 → 1.1.1) replaces the links-only rule with a broader "question carries what it asks about" rule: Step 7 puts the full plan in Looks good preview; Step 6/8 flag-release steps follow release-plans; Step 1 and test-PR flows embed summary/diff in questions.

Evals add runQuestionShowsCommand (curl in Run-this-command questions) and confirmShowsPlan (list plan naming every repo written), wired into ~20 promptfoo cases, plus unit tests. skills.json / marketplace metadata pick up version bumps.

^{Reviewed by Cursor Bugbot for commit d08fa7a. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

Skill eval results

Skill	Before	After	Δ
agentcontrol/configs-create	100/100 (4/4)	100/100 (4/4)	no change
agentcontrol/configs-update	80/100 (4/5)	80/100 (4/5)	no change
agentcontrol/configs-variations	80/100 (4/5)	80/100 (4/5)	no change
agentcontrol/tools	75/100 (3/4)	75/100 (3/4)	no change
aws-devops-agent/experiment-orchestration	-	100/100 (9/9)	new
experiments/launchdarkly-experiment-guide	-	100/100 (5/5)	new
experiments/launchdarkly-experiment-hypothesis-builder	-	100/100 (13/13)	new
experiments/launchdarkly-experiment-setup	-	100/100 (8/8)	new
feature-flags/flag-and-release-change	-	100/100 (5/5)	new
feature-flags/flag-release	-	86/100 (6/7)	new
feature-flags/launchdarkly-flag-command	-	100/100 (3/3)	new
feature-flags/launchdarkly-flag-create	100/100 (3/3)	100/100 (4/4)	no change
feature-flags/launchdarkly-flag-drift	-	100/100 (4/4)	new
feature-flags/should-flag-change	-	100/100 (17/17)	new
onboarding	-	86/100 (6/7)	new
release-autopilot/launchdarkly-release-autopilot-onboarding	-	79/100 (30/38)	new
release-autopilot/launchdarkly-release-plans	-	88/100 (15/17)	new

Only suites whose source actually changed since their last recorded score were re-run. Soft-failing while we stabilise the baseline.