[JacobStephens2]

Summary

- Run / Merge run / Failed run
+ Ticket run / optional Self-merge / Failed ticket run
- Weeding / Fencing
+ Architect runs every few minutes / Security runs on a schedule, with fixing allowed

Renames the glossary, agent-facing text, manual and README together; regenerates prompts and site pages. ADR 0022 records why stored Run kinds remain stable. Old README bookmarks and manual links still resolve.

Closes #1102

Evidence

• Before: node --test --test-name-pattern='manual starts folded' site/tests/pages.test.cjs failed looking for “What a Ticket run does”.
  After: the same navigation test passes at 320px without JavaScript.
• Prompt regeneration and cargo test prompts_page: 12 passed. cargo fmt --check: passed.
• scripts/affected-tests issue-1101 selected the browser and full Rust suites: 45 site tests and 1,243 active unit tests passed. It stopped at a host-only systemd warning from the installed snapd.service. With a temporary process-scoped Snapd mask, all 7 site_deploy tests passed; the seven remaining integration binaries then passed. Coverage reconciliation against Cargo metadata confirms all 56 integration binaries passed.
• Host overrides: per-process tag.gpgsign=false prevents this machine's signed-tag prompt; SYSTEMD_UNIT_PATH isolates its incompatible Snapd unit. No repository or host-service changes were made for these environment issues. Timed selector: 1,020.00s; targeted continuation: 281.09s. Incremental compilation was included; dependency compilation ran earlier.
• Both reviewers' notification checks failed before the correction; the retained notification documentation limits suppression to child Ticket runs test now passes.

Test seams

• Rendered manual navigation and repository links, through the existing site tests.
• Notification ownership prose in the glossary, manual and rendered manual, through the retained review regression test.
• Generated Session prompts and embedded Factory skills, through prompts_page and site browser tests.

Review coverage

Independent Standards and Spec reviews compare against issue-1101. Each read all 25 changed files, including generated output and the final CI policy wording. Both identified the notification child-qualifier defect; their exact reproductions failed, and the correction and regression test were reviewed. Standards: 1 finding resolved. Spec: 1 finding resolved.

Unread changed files: none on either axis. Generated-output exclusions: none. Full reports and each axis's final files-read list are saved in .thirdshift-review-5b6VQm/standards.md and .thirdshift-review-5b6VQm/spec.md in the worktree.

Unaddressed findings

Standards

None skipped; the sole finding was reproduced and fixed.

Spec

None skipped; the sole finding was reproduced and fixed.

Merge Danger

Door: two-way

Blast Radius: documentation

The change affects agent instructions, generated documentation and manual anchors. CLI output and stored kinds retain their current spelling for the next Ticket.

Built with codex · gpt-6.1-sol · high