design-kanban-md-outpu…
Preserve and evolve kanban-md table, compact, and JSON output contracts. Use when changing…
Plan and run verification for kanban-md changes, including test-driven bug fixes, Go package tests, CLI end-to-end tests, TUI behavior and snapshots, lint, and precommit checks. Use when implementing or reviewing a change, fixing a bug, adding tests, updating snapshots, or
$ npx -y skills add antopolskiy/kanban-md --skill test-kanban-md --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/test-kanban-mdContext preview
The summary Claude sees to decide when to auto-load this skill.
Plan and run verification for kanban-md changes, including test-driven bug fixes, Go package tests, CLI end-to-end tests, TUI behavior and snapshots, lint, and precommit checks. Use when implementing or reviewing a change, fixing a bug, adding tests, updating snapshots, or
name: test-kanban-md description: > Plan and run verification for kanban-md changes, including test-driven bug fixes, Go package tests, CLI end-to-end tests, TUI behavior and snapshots, lint, and precommit checks. Use when implementing or reviewing a change, fixing a bug, adding tests, updating snapshots, or deciding what evidence is required. Do not use as the source of release or schema-migration policy. allowed-tools: - Bash(go *) - Bash(golangci-lint *) - Bash(make *) - Bash(git diff *)
Choose the smallest fast feedback loop while developing, then widen verification in proportion to the change before declaring it complete.
test and confirm it fails for the expected reason, then implement the fix.
command tests for CLI wiring, E2E tests for complete binary workflows, and TUI behavioral or snapshot tests for rendered interaction.
change, and the full suite before completing production-code work.
inspect the resulting diff rather than accepting it blindly.
failures, and checks that were not run.
1. Identify the affected package or interface and study nearby test patterns. 2. Add the narrowest test that reproduces the reported behavior. 3. Run that test and confirm the pre-fix failure is meaningful. 4. Implement the smallest sufficient fix. 5. Run the reproducer again and confirm it passes. 6. Run the affected package or E2E suite. 7. Run the full repository tests and lint. 8. Review snapshots, generated files, and the final diff for unintended changes.
Do not weaken an assertion merely to make a failing test pass. If the expected behavior is unclear, resolve that product decision before encoding it.
# One test or package during development go test ./path/to/package -run TestName -count=1 go test ./path/to/package -count=1 # Repository-wide checks go test ./... golangci-lint run ./... # Full branch validation used by the repository make precommit
Use `make precommit` before integration when the change warrants the full build, race-enabled coverage, E2E, lint, and clean-worktree checks. If it fails because of a pre-existing baseline, verify the current diff separately and report the baseline precisely; do not silently present a partial check as a full pass.
For package locations, helpers, and TUI snapshot commands, read [references/test-patterns.md](references/test-patterns.md).
An agent-first, file-based Kanban board for coordinating AI coding agents and human supervisors. It runs locally as a single binary: no database, server, account, or SaaS dependency.
Preserve and evolve kanban-md table, compact, and JSON output contracts. Use when changing…
Evolve kanban-md config.yml schemas and task Markdown frontmatter without breaking existing…
Autonomous, parallel-safe development workflow using kanban-md. Use when the user asks to…
Review kanban-md feature requests, issues, PRs, and design proposals for product fit,…
Manage project tasks using kanban-md, a file-based kanban board CLI. Use when the user…
Release kanban-md through its tag-triggered GoReleaser workflow, monitor CI, recover safely…