Skip to main content

ccr-gate-recovery

CCR-specific local gate recovery workflow. Use when asked to fix failing CCR checks, just ci, Rust workspace gates, ccr-ui checks, VS Code extension tests, docs build/audit failures, or noisy aggregate CI output. Applies to Claude Code, Codex, Grok, Kimi, and OMP. Keep existing parallel tests; do not invent a new recovery engine.

Source facts

Repository
bahayonghang/ccr
Last source activity
October 3, 2026 at 15:20
Detected SKILL.md language
English
Stars
3
Forks
0

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
ccr-gate-recovery
description
CCR-specific local gate recovery workflow. Use when asked to fix failing CCR checks, just ci, Rust workspace gates, ccr-ui checks, VS Code extension tests, docs build/audit failures, or noisy aggregate CI output. Applies to Claude Code, Codex, Grok, Kimi, and OMP. Keep existing parallel tests; do not invent a new recovery engine.
# CCR Gate Recovery Recover local verification gates from the CCR repository root by finding the first real failing step, fixing narrowly, and escalating validation only after the narrow gate is green. This skill applies to Claude Code, Codex, Grok Build, Kimi Code, and OMP. Use existing repository gates and default test parallelism. Isolate shared environment state in the owning fixture. Keep binding exports in the separate generation gate. Before implementation approval, inspect and run only authorized checks. After approval, implement and self-fix check roles may edit the approved files. Keep native client, hosted CI, and local test evidence separate. ## Workflow 1. Confirm the current working tree with `git status --short`; preserve unrelated user changes. 2. Reproduce the narrowest failing gate when the user gave one. For aggregate output, identify the first real failure and ignore downstream noise until it is fixed. 3. Inspect only the owning surface: - Rust workspace: `crates/**`, `Cargo.toml`, root `justfile`. - UI/Tauri: `ccr-ui/src/**`, `ccr-ui/src-tauri/**`, `ccr-ui/tests/**`. - VS Code extension: `ccr-vscode/src/**`, `ccr-vscode/package.json`. - Docs: `docs/**`, especially `.vitepress/config.mjs` and `scripts/audit-docs.mjs`. 4. Make the smallest behavior-preserving or bug-fixing edit. Do not add dependencies unless explicitly requested. 5. Re-run the same failing gate. When green, run the next broader gate that proves the touched surface. 6. Finish with `git diff --check` and report exact commands run plus any skipped expensive gates. ## Gate Map ### Rust ```powershell cargo fmt --all --check cargo clippy --workspace --all-targets --all-features -- -D warnings cargo test --workspace --all-features -- --skip export_bindings ``` Use package-scoped `cargo test -p <crate-name> -- --skip export_bindings` when the failing crate is clear. Independent UI, docs, and extension gates may run in parallel with Rust. ### ccr-ui ```powershell cd ccr-ui bun run type-check bun run lint bun run test ``` Use root `just ui-check` for the broader UI gate and `just frontend-check` when docs also need to be included. ### VS Code extension ```powershell cd ccr-vscode npm run build npm test ``` Use `npm run lint` or `just ci` inside `ccr-vscode/` when TypeScript config or packaging surfaces changed. ### Docs ```powershell cd docs bun run build bun run audit ``` Fix documentation audits at the source Markdown/config level; do not edit `.vitepress/dist/`. ### Full repository ```powershell just ci ``` Run this only after narrow gates are green or when the user explicitly asks for the full repo gate. ## Reporting Keep the final report short: - failing gate and root cause - files changed - validation commands and pass/fail status - remaining risks or skipped broader gates
View on GitHub