This skill provides post-deploy canary monitoring: it visits specified pages, captures screenshots, records console errors and performance metrics, compares results to a baseline, and alerts on regressions or new failures. It's built as a safety net that watches production immediately after deployment.
Invoke this after a deployment to verify production health (default 10-minute monitoring). Use --baseline to capture pre-deploy references, or run quick checks for single pages. Ideal for release verification, smoke testing, and fast incident detection.
Inference: Designed for Claude Code / ostack tooling and any agent that can call the browse daemon or run Bash steps to snapshot pages and collect diagnostics.
Post-deploy canary monitoring skill that uses a browse daemon to watch live apps for console errors, performance regressions, and page failures with periodic screenshots and baseline comparisons. The SKILL.md is detailed and well-structured with clear phases, but the preamble contains auto-update mechanisms (ostack-update-check, ostack-upgrade flow) and sources external shell scripts (ostack-repo-mode) which are security concerns. No bundled scripts to test. The skill depends heavily on ostack ecosystem tooling (browse daemon, ostack-slug, ostack-config) that must be pre-installed.
Well-written monitoring skill with clear phases and good alerting logic (transient tolerance, baseline comparisons). Main security concerns: preamble evals external scripts, curl|bash for bun install, and auto-update mechanism. The ostack ecosystem dependency means the skill is not standalone — it requires significant setup. The SKILL.md is verbose with lots of ostack-internal philosophy (Conviction and Completeness, Repo Ownership Mode, Contributor Mode) that bloats the document and dilutes the canary-specific instructions.