Track first 10 real-world AgentLayer alpha scans

[Qqqq5910]

This issue tracks the public-alpha feedback target for AgentLayer's first month.

Goal

Collect real public-site scan feedback from early testers and turn it into actionable heuristic/report improvements.

Targets

• 10 real public-site scans submitted or shared safely
• 5 actionable feedback reports triaged
• 5 high-frequency false-positive or confusing-recommendation patterns identified
• 2 high-frequency issues fixed or converted into regression tests
• Follow-up release notes or changelog entry written for fixes learned from feedback

How testers can contribute

Ask testers to run:

pnpm dlx @junyi5910/agentlayer-cli generate https://your-site.com --out ./agentlayer-output --max-pages 20

Then comment on the pinned feedback issue:

#1

Safety

Only public URLs and non-sensitive excerpts should be shared. Do not include secrets, credentials, private URLs, internal paths, customer data, confidential screenshots, unpublished launch plans, cookies, tokens, or form submission data.

[Qqqq5910]

Released v0.2.0-alpha.2 to fix generated report layout overflow. Wide task/result tables now stay inside the viewport with internal horizontal scrolling, and long URLs/snippets wrap instead of stretching the whole page.\n\nVerified:\n- npm latest/alpha tags point to @junyi5910/agentlayer-cli@0.2.0-alpha.2\n- CI passed on main\n- Published CLI smoke passed\n- Fresh pnpm dlx install generated a Linear report without page-level horizontal overflow on desktop and mobile viewports

[Qqqq5910]

Month-one status update after v0.2.0-alpha.3:\n\nCompleted:\n- npm alpha install path works: @junyi5910/agentlayer-cli latest/alpha -> 0.2.0-alpha.3\n- README and release metadata point to v0.2.0-alpha.3\n- CI passed on main for commit 9bbdb7b\n- Published CLI smoke passed after npm publish\n- Generated report layout overflow fixed in v0.2.0-alpha.2\n- Fragment-only form actions such as action="#" fixed in v0.2.0-alpha.3\n- examples/real-world now has 20 anonymized public-site scan summaries\n- docs/heuristics-known-issues includes real-scan patterns for canonical-domain redirects and fragment-only form actions\n\nStill needs external input to fully satisfy the month-one feedback goal:\n- 10 real testers or site owners invited\n- 5 actionable external feedback reports collected\n\nPinned feedback issue for invitees: #1

[Qqqq5910]

Month-one launch materials are now committed on main in c71e801. Added docs/launch/tester-kit.md for sending to invited testers, docs/launch/month-one-audit.md to separate completed repo/release work from still-pending external feedback, and pinned tester-facing scan commands to v0.2.0-alpha.3. CI passed in run 27473536534. Remaining non-code work: invite real testers, collect at least 5 actionable external reports, and convert repeatable issues into GitHub issues.

[Qqqq5910]

Added launch feedback operations docs in 5229c13. This includes docs/launch/README.md, docs/launch/outreach-message-pack.md, docs/launch/feedback-intake.md, and maintainer simulation rows in docs/launch/outreach-tracker.md for PostHog, Clerk, Resend, Vercel, and Neon. The simulation rows are explicitly marked as seed examples and do not count as external-user evidence. CI passed in run 27473738624. Remaining completion evidence still requires real external testers to submit scan results.

[Qqqq5910]

Converted two maintainer simulation findings into labeled candidate issues: #3 for canonical-domain redirect coverage diagnostics and #4 for high-signal bounded crawl prioritization. Added the maintainer-seed label to both so they are not confused with external-user feedback. Updated docs/launch/outreach-tracker.md and docs/launch/month-one-audit.md in 78ae9b1. CI passed in run 27473885735. These are useful seed issues, but the month-one external feedback target still requires real testers to submit scan results.

[Qqqq5910]

Follow-up from maintainer-seed feedback: c3b07e2 fixed #3 and #4. The scanner now surfaces canonical-domain redirect coverage as a first-class recommendation and prioritizes high-signal task pages in bounded scans, with regression tests. CI passed in run 27474173843. This improves the first-month scan experience, but external-user feedback is still the remaining completion evidence.

[Qqqq5910]

Added China social launch drafts in 6b409c8: docs/launch/xiaohongshu-post.md and docs/launch/douyin-post.md. The launch README now links them, launch-posts commands are pinned to v0.2.0-alpha.3, and the outreach tracker includes Xiaohongshu/Douyin as channels. CI passed in run 27474327040. These materials make external tester acquisition easier; completion still requires real users to submit scan feedback.

[Qqqq5910]

Removed platform-specific Xiaohongshu/Douyin launch drafts from the open-source repository in c2d6466 after deciding they should stay outside the public project repo. Cleared all related links/references from docs/launch. CI passed in run 27474455600.