|
|
@@ -22,10 +22,11 @@ From Python async patterns to Rust ownership models, from AWS Fargate deployment
|
|
|
|
|
|
## Recent Updates
|
|
|
|
|
|
-**v2.5.1** (May 2026)
|
|
|
-- 🩺 **`windows-ops` skill** - Comprehensive Windows workstation operations rendered as a proper claude-mods instrument. Brand `🩺` (stethoscope) — diagnostics is the verb. Seven scripts (`health-audit`, `disk-health`, `crash-triage`, `drive-dependencies`, `recover-clone`, `safe-disable-startup`, `boot-perf`) plus five reference catalogs (storage events, BugCheck codes, startup mechanisms, recovery patterns, remote diagnostics). The killer feature is `health-audit.ps1` which renders a state-grouped panel: failing/warn/pass/info sections, critical alert with cross-script wayfinding when a failing drive is detected, footer indicators showing high-action signals (`⬤ storage` busted + `• N crashes`). Maps `\Device\HarddiskN` → drive letter so the verdict identifies a specific drive, not a Disk number. `crash-triage.ps1` decodes Event 41 with `Properties[0]` (the actual BugCheck code, not `Properties[1]` like most Stack Overflow answers) and walks the N minutes before crash flagging smoking guns. `safe-disable-startup.ps1` implements the StartupApproved registry mechanism so non-admin users can disable HKLM startup entries. `drive-dependencies.ps1` is the "safe to disconnect?" check before pulling a drive. `recover-clone.ps1` uses `robocopy /R:0` so retries don't accelerate a failing drive's death. Dogfooded on a real workstation with a dying 8TB HGST HDD — correctly identified the drive (1943 Event 7 + 1646 Event 154 + 20 storahci resets), decoded two recent crashes with cause discrimination (PowerButtonTimestamp non-zero = forced shutdown of hung machine; zero = hard power loss). Bug exposed and fixed during the visual refactor: storahci resets had been attributed to every disk's verdict, falsely flagging healthy drives as FAILING.
|
|
|
-- 🎨 **`_lib/term.ps1`** - PowerShell port of the Terminal Panel Design System (`docs/TERMINAL-DESIGN.md`). Mirror of `_lib/term.sh` with PowerShell-native idioms (Verb-Noun cmdlets, `$Script:` scope, `ValidateSet` params, `Get-TermDisplayWidth` for emoji-aware width handling). Gives PowerShell-based skills the same panel chrome, glyph palette, color tokens, and grid as bash siblings. Three implementations of the spec now coexist: `term.sh` (fleet-ops + future bash), `term.ps1` (windows-ops + future PowerShell), and the `Term` class in `summon.py` (summon + future Python). Brand registry gains `windows-ops = 🩺 | [H]`. The design doc's § 9 section renamed from "Implementation — `skills/_lib/term.sh`" to "three siblings of the same spec" acknowledging the multi-language reality. All seven windows-ops scripts source `term.ps1` and render panels; JSON output paths remain pure data with no chrome contamination.
|
|
|
+**v2.6.0** (May 2026)
|
|
|
+- 🩺 **`windows-ops` skill** - Comprehensive Windows workstation operations rendered as a proper claude-mods instrument. Brand `🩺` (stethoscope) — diagnostics is the verb. Seven scripts (`health-audit`, `disk-health`, `crash-triage`, `drive-dependencies`, `recover-clone`, `safe-disable-startup`, `boot-perf`) plus five reference catalogs (storage events, BugCheck codes, startup mechanisms, recovery patterns, remote diagnostics). The killer feature is `health-audit.ps1` which renders a state-grouped panel: failing/warn/pass/info sections, critical alert with cross-script wayfinding when a failing drive is detected, footer indicators showing high-action signals (`⬤ storage` busted + `• N crashes`). Maps `\Device\HarddiskN` → drive letter so the verdict identifies a specific drive, not a Disk number. `crash-triage.ps1` decodes Event 41 with `Properties[0]` (the actual BugCheck code, not `Properties[1]` like most Stack Overflow answers) and walks the N minutes before crash flagging smoking guns. `safe-disable-startup.ps1` implements the StartupApproved registry mechanism so non-admin users can disable HKLM startup entries. `drive-dependencies.ps1` is the "safe to disconnect?" check before pulling a drive. `recover-clone.ps1` uses `robocopy /R:0` so retries don't accelerate a failing drive's death. `boot-perf.ps1` measures boot duration from the Diagnostics-Performance log (admin) or kernel events (non-admin), flagging the slowest boot in window with a capacity pip bar. Dogfooded on a real workstation with a dying 8TB HGST HDD — correctly identified the drive (1,943 Event 7 + 1,646 Event 154 + 20 storahci resets), decoded two recent crashes with cause discrimination (PowerButtonTimestamp non-zero = forced shutdown of hung machine; zero = hard power loss). Bug exposed and fixed during the visual refactor: storahci resets had been attributed to every disk's verdict, falsely flagging healthy drives as FAILING.
|
|
|
+- 🎨 **`_lib/term.ps1`** - PowerShell port of the [Terminal Panel Design System](docs/TERMINAL-DESIGN.md). Mirror of `_lib/term.sh` with PowerShell-native idioms (Verb-Noun cmdlets, `$Script:` scope, `ValidateSet` params, `Get-TermDisplayWidth` for emoji-aware width handling). Gives PowerShell-based skills the same panel chrome, glyph palette, color tokens, and grid as bash siblings. Three implementations of the spec now coexist: `term.sh` (fleet-ops + future bash), `term.ps1` (windows-ops + future PowerShell), and the `Term` class in `summon.py` (summon + future Python). Brand registry gains `windows-ops = 🩺 | [H]`. The design doc's § 9 section renamed from "Implementation — `skills/_lib/term.sh`" to "three siblings of the same spec" acknowledging the multi-language reality. All seven windows-ops scripts source `term.ps1` and render panels; JSON output paths remain pure data with no chrome contamination.
|
|
|
- 📐 **`docs/DESIGN.md` → `docs/TERMINAL-DESIGN.md`** - Renamed for specificity. The doc was always about terminal panel rendering; the generic name was a collision risk. All 6 cross-references updated (README, rules, term.sh, fleet.sh, summon SKILL.md + summon.py).
|
|
|
+- 🔧 **Diagnostic-script exit-code semantics** - `health-audit`, `disk-health`, `drive-dependencies` now always exit 0 on successful run regardless of findings. Critical issues surface in the panel output (red sections, `▲` alerts, `⬤` busted footer indicator) and JSON, not via non-zero exit codes. Eliminates false "Background task failed" labels in UIs that treat any non-zero exit as broken. Automation should branch on the JSON `verdict` field, not `$LASTEXITCODE`.
|
|
|
|
|
|
**v2.5.0** (May 2026)
|
|
|
- 🌐 **`net-ops` skill** - Cross-platform network troubleshooting (Windows / macOS / Linux) via local or remote SSH with a layered diagnostic ladder: link → ICMP → socket → DNS infrastructure → OS resolver → app. NDP-aware IPv6 classifier (disabled / ULA-only / no-route / path-broken / healthy), MTU/PMTU test, time-skew check, browser DoH detection (Chrome / Brave / Firefox), WSL2/container awareness. Modes: `--watch`, `--json` (NDJSON), `--redact` for opsec-clean dumps, `--quick` for skip-if-healthy. Per-OS probe + dns-audit + repair scripts, reverse-mode probe, 24-test self-suite.
|