Skip to content

docs(skill): simplify first-run sign-in flow to single onboarding prompt - #132

Merged
yunze7373 merged 2 commits into
mainfrom
docs/skill-first-run-one-question
Oct 2, 2026
Merged

yunze7373 merged 2 commits into
mainfrom
docs/skill-first-run-one-question

Conversation

@yunze7373

Copy link
Copy Markdown
Member

Summary

Streamline the XMemo Skill first-run onboarding sequence in SKILL.md:

  • Single Onboarding Prompt: Use the canonical template ("XMemo is your personal cloud memory ... Sign in to get started?") as the sole initial prompt. Remove the redundant unencrypted-storage disclosure prompt and example sentence.
  • Non-blocking Device Login: Guide agents to run login --allow-plaintext in a background or streaming shell so the verification URL and code appear immediately without blocking on user action or timing out.
  • One-line Connection Confirmation: Report connection in one line following auth status --verify.
  • Silent Re-offer Initialization: Remove the inline AGENTS.md offer prompt at sign-in. Instead, the agent silently runs profile --status later so subsequent recall-based re-offers activate per existing runtime logic.
  • Diagnostics Reordered: Put the sign-in sequence first under ## First Successful Run, relocating diagnostic checks (doctor --anonymous, XMEMO_KEY, auth status --verify) to a dedicated ### Diagnostics and Advanced Setup section.

Verification

  • node --test test/xmemo-skill.test.js test/xmemo-skill-snapshot.test.js: 24/24 pass
  • node scripts/verify-release-packaging.mjs: 4/4 drills pass
  • npm run lint: 0 errors
  • npm test: 651 tests, 650 pass, 1 skipped, 0 fail
  • SkillSpector static scan: 3 issues (matching baseline; 0 PE3 findings in markdown files)
  • Zero occurrences of Continue with browser sign-in or separately disclose across skills/

pc-gemini added 2 commits October 2, 2026 17:35
Streamline the XMemo Skill first-run onboarding sequence in SKILL.md:
- Merge initial sign-in into a single canonical prompt and browser approval
- Remove separate unencrypted-storage disclosure prompt and inline AGENTS.md offer
- Instruct agents to execute login in a non-blocking shell, outputting only the verification URL, code, and approval prompt
- Silently initialize profile status to 'later' upon successful connection for subsequent recall-driven re-offers
- Place the sign-in flow first, moving diagnostic and secret-store setup commands into an advanced section
- Add unreleased changelog entry and update test assertions
Add the --if-unset flag to the profile command to prevent first-run sign-in from overwriting an existing user status choice (e.g. 'never'). When --if-unset is passed and a status is already stored, leave the state unchanged and exit 0. Update SKILL.md step 3 to use the guarded flag, add test cases for unset, never, and max-offers conditions, update help snapshots, and note the change in CHANGELOG.
@yunze7373
yunze7373 merged commit 0d3a00c into main Oct 2, 2026
7 checks passed
@yunze7373
yunze7373 deleted the docs/skill-first-run-one-question branch October 2, 2026 09:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant