[csd113]

Problem and resulting behavior

The README showed only one desktop launcher image and buried the platform and installation boundaries. Lead with what Vitrallis does, add a vertical tour of the launcher, Terminal, Notepad, Files, Settings and App Center, and make setup, controls, limits and detailed guides easier to find.

All six new images are unaltered 480×272 captures from a physical PocketCHIP running the verified published 1.0.4 binaries. The launcher and Settings use live battery/Wi-Fi status; native apps use harmless temporary sample files. App Center shows its own interface before refresh, without third-party apps. Record bundle/source provenance and the bounded capture scope in docs/images/README.md.

App Center’s concurrent-install test could wait forever when the service failed before starting a download, concealing other CI failures. Bound the test gate waits, disconnect the start signal before joining, and include the service updates in an installer error. Existing concurrency assertions and production behavior are preserved. Run the unchanged Linux CI gates as the unprivileged runner user in a private PID/procfs namespace: unrelated runner processes can deny executable inspection despite sharing the runner UID, and App Center deliberately fails closed on that uncertainty. All test fixture processes remain visible; no tests or permission checks are skipped. Set umask 077 inside validation so filesystem fixtures cannot inherit the runner’s group-writable creation mask.

Validation

Passed locally on the macOS development host:

• cargo fmt --all --check
• cargo clippy --workspace --all-targets --all-features -- -D warnings -D clippy::all -D clippy::pedantic -D clippy::nursery -D clippy::cargo
• cargo test --workspace --all-features — 389 passed; 9 explicitly ignored hardware/online checks.
• sh scripts/validate.sh — workspace checks, Rust/Python suites, release builds, native renderer and SDL smoke checks, script syntax, documentation links and diff whitespace. Python discovery ran 181 tests with 9 explicit skips; the separate native-renderer pass ran 2 with 1 skip.
• python3 scripts/check-doc-links.py — 62 Markdown files.
• All six image references exist, have descriptive alt text, decode as 480×272 PNGs and match the device-side SHA-256 hashes after transfer.
• Reviewed the GitHub Markdown API render with GitHub-style CSS at 1100px desktop and 390px mobile widths in Chromium: all images render, no page overflow, commands remain selectable and scroll within their blocks.
• The concurrent-install and mid-fetch lock tests pass. A Linux diagnostic that injects an early process-inspection error now reports that error in 0.08 seconds instead of hanging.
• sh scripts/validate.sh also passes on Linux ARM in the same private PID/procfs namespace, as an unprivileged user with umask 077. This includes all workspace/Python tests, release builds, native-renderer and SDL smoke checks, documentation links and diff checks.
• Workflow YAML parses and the complete final diff was reviewed.
• All four GitHub checks passed on f4c89ac: push and PR matrices for Rust 1.99.0 and stable.

Scope and remaining risks

README, capture provenance, shell/validation image references, six lossless PNGs (about 35 KiB total), the explicitly authorized App Center test-fixture deadlock fix, and CI process-namespace isolation. The installation command and entire uninstall section are preserved exactly. Application behavior, dependencies, versions, historical images and existing user changes are unchanged.

The PocketCHIP's installed 1.0.3 session and data were preserved; 1.0.4 ran only from a private temporary capture workspace, which was removed afterward. This capture session is not full 1.0.4 certification and does not retest installation, cold boot, physical input, audio or storage durability. No release or Reddit post is created.