Pandahrms skills for Claude Code: a manual, standalone work flow -- /discover, /discover-ticket, and /discover-project intake, /spec (layered Gherkin behaviour spec), /slice (vertical-slice cards), /execute (guided card run with TDD and stop-gates), /status, /close (which invokes /resolve-ticket for ticket work), /resolve-ticket (card-less ticket resolution), /pr -- plus a decomposed review/verify/commit family split by scope and job: /lint-gate (diff-scoped deterministic guard runner -- linter, scans, analyzers, L1->L2 traceability), /verify (project-scoped runner -- full build + full test + coverage, the single source /commit and /execute invoke), /code-review (diff-scoped LLM judgment), /pr-approver-review (senior-approver review of an already-opened GitHub PR by number, cross-checking claude[bot]), /security-review, /simplify, /card-commit (card-scope commit trusting the card's pre-complete /verify), and /commit (branch-scope commit gate invoking /verify) -- plus safe branching, EF Core migrations, handoff-compact, tool-doctor, and the optimise-prompt pre-flight repeat-back
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
ALWAYS use when the user needs to create, start, or checkout a new git branch. Prevents dangerous upstream tracking inheritance and enforces folder-based naming conventions. Triggers on any mention of new branch, new feature branch, branching off, or checkout -b
Manually invoked as `/card-commit` (or by an explicit mention of "card-commit" / "commit this card") to commit ONLY the current card's slice of changed files -- not the whole branch. Trusts the card's pre-complete `/verify` already returned `VERIFY RESULT: PASS` on this tree with no edits since: runs format + lint, but does NOT re-run the test suite or build. If any edit landed after that PASS (or the PASS cannot be confirmed), invokes `/verify` first and requires PASS before committing. Plans and executes atomic commits of the card's files with a conventional message and no AI attribution. Never pushes, never opens a PR, never amends or runs destructive git. Does NOT auto-trigger -- only on the `/card-commit` slash command or an explicit "card-commit"/"commit this card" mention; a finished card alone is not enough.
Manually invoked as `/close` to close a finished piece of Pandahrms work after every card is executed. Verifies every card is already in the per-work `done/` folder and STOPS if any card remains in `active/`; for ticket-driven work invokes `/resolve-ticket` to move the ticket to a resolved, ready-for-release state (status + solution, dev status, Developer Resolution, customer comment); writes the docspace dev-diary log and updates progress; marks the work closed in the per-work `_overview`. Mutating -- it changes ticket state and writes the log. Does NOT move cards. Does NOT raise PRs (that is `/pr`). Does NOT auto-trigger -- only on the slash command or an explicit "close this work" mention.
Triggers on mentions of code review of working-tree changes -- "review my changes", "check my changes", "review the diff", "review before commit". If the user is ambiguous between working-tree review and PR review, ask which one they mean. A diff-scoped LLM-JUDGMENT review only -- SOLID intent, naming meaning, semantic reuse, audit-pattern conformance, input-validation adequacy, PII/data-exposure, error-handling adequacy, readability, and spec MEANING. Reads all changed files, reviews against the judgment checklist, fixes issues, dispatches a Codex second opinion when the diff warrants it, and runs /simplify. Does NOT run the linter, build, tests, or deterministic guards -- those belong to /lint-gate and /verify; when a /lint-gate result is provided, code-review consumes its `OWNED:` line and drops the matching judgment rows, merging its `[tool:*]` findings as context. Does NOT commit directly. Supports `--approve` to auto-pick the "proceed" choice at every AskUserQuestion, and `--skip` to forward to `/commit --skip` at Phase 7. `--codex` / `--no-codex` force the Codex second opinion on or off.
Triggers when the user explicitly requests a git commit of the whole branch's working-tree changes -- phrases like "commit my changes", "git commit", "ready to commit", "/commit", or "make atomic commits". Branch-scope commit step: a clean-tree gate over the whole branch that auto-fixes format + lint inline, then invokes `/verify` (the project-scoped build + test runner) and requires `VERIFY RESULT: PASS` before planning and executing atomic commits across the branch. The gate applies even to pre-existing or unrelated failures. Supports `/commit --skip` to bypass the entire Phase 1 gate (format, lint, and the `/verify` build + test run).
Pandahrms-specific skills plugin for Claude Code.
The plugin is a set of manual, standalone skills. There is no orchestrator -- you run each step yourself.
| Skill | Slash Command | Description |
|---|---|---|
| discover | /pandahrms:discover | Free-form intake door: a new feature / enhancement / bug -> objective + acceptance criteria |
| discover-ticket | /pandahrms:discover-ticket | Ticket intake door (workspace-prod MCP) -> same output contract as /discover |
| discover-project | /pandahrms:discover-project | Project-queue door (workspace-prod MCP) -> numbered table of a project's pending dev tickets; user picks -> /discover-ticket |
| spec | /pandahrms:spec | Write/update the L1 behaviour Gherkin spec in pandahrms-spec; conditional on behaviour change; user-agreement gate |
| slice | /pandahrms:slice | Cut agreed work into vertical-slice cards; each card holds its L2 spec files + an ordered work sequence |
| execute | /pandahrms:execute | Run one card: guided run with stop-gates, spec-first TDD, inline review/deploy/regen; /execute card-NN or bare /execute for the next card |
| status | /pandahrms:status | Read-only summary: auto-fires when /execute finishes the last card, also a manual status report |
| close | /pandahrms:close | Mutating close: re-check all cards done, update ticket status, write the log, tidy cards |
| pr | /pandahrms:pr | Optional final PR: runs /commit first, then raises the PR (ticket reference in the body) |
/execute; also standalone)| Skill | Slash Command | Description |
|---|---|---|
| code-review | /pandahrms:code-review | Review git changes against code standards, fix issues, run /simplify (no commits) |
| security-review | /pandahrms:security-review | Security review (OWASP Top 10 + Pandahrms tenant/audit/PII checks) (no commits) |
| simplify | /pandahrms:simplify | 3-agent reuse/quality/efficiency pass on working-tree changes (no commits) |
| commit | /pandahrms:commit | Verify the working tree is clean, plan and execute atomic commits |
| Skill | Slash Command | Description |
|---|---|---|
| optimise-prompt | (runs each turn) | Repeat-back in B1 English (keep technical terms) so you can confirm Claude read you right |
| branching | /pandahrms:branching | Safe branch creation with upstream protection and folder-based naming |
| ef-migrations | /pandahrms:ef-migrations | EF Core migration commands for Performance and Recruitment APIs |
| handoff-compact | /pandahrms:handoff-compact | Write a session handoff doc, then compact |
/plugins marketplace add https://github.com/pandaworks-software-plt/pandahrms-skills.git
/plugins install pandahrms@pandahrms-skills
To update the plugin to the latest version:
/plugins update pandahrms@pandahrms-skills
Each skill is manual and standalone -- you run each step. There is no orchestrator: each skill ends by suggesting the next one (## Next step), so the flow self-guides while you stay in control of every hop. A typical piece of work flows like this:
/discover (or /discover-ticket; /discover-project to pick from a project's pending queue) intake -> objective + acceptance criteria
-> /spec L1 behaviour Gherkin spec (central pandahrms-spec)
-> /slice vertical-slice cards (each holds its L2 spec files + an ordered sequence)
-> /execute per card: guided run with stop-gates + spec-first TDD
inline leaf actions: /code-review (+ /security-review when sensitive),
deploy BE, regen FE types, per-card commit/PR
-> /status auto when the last card is done (conclusion) + manual status anytime
-> /close update ticket status, write the log, tidy cards
-> /pr optional final PR (runs /commit first)
The always-on execution rules (TDD markers, gates, sensitivity list) ship via the plugin's SessionStart hook, so they apply in every session with no per-member setup.
UNLICENSED - Proprietary. Pandaworks Software Plt.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimnpx claudepluginhub pandaworks-software-plt/pandahrms-skills --plugin pandahrmsComplete collection of battle-tested Claude Code configs from an Anthropic hackathon winner - agents, skills, hooks, and rules evolved over 10+ months of intensive daily use
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Comprehensive .NET development skills for modern C#, ASP.NET, MAUI, Blazor, Aspire, EF Core, Native AOT, testing, security, performance optimization, CI/CD, and cloud-native applications
Next.js development expertise with skills for App Router, Server Components, Route Handlers, Server Actions, and authentication patterns
Unity Development Toolkit - Expert agents for scripting/refactoring/optimization, script templates, and Agent Skills for Unity C# development