sgd

Plugin sgd from marketplace wtp-plugins in WealthTechPros/sge-public. Commands read from the repo's manifest — today.

Skills (51)

• agent-template

  Reference template for custom orchestrators that dispatch eligible issues and run fresh issue-bound governance classification before writes.

• atomic-audit

  Use when scoring how far a repo or portfolio has adopted atomic design — design tokens, primitive layer, catalog, enforcement maturity — for /sge:sge-align check C10 or before planning a design-system roadmap. Advisory and read-only.

• available-issues

  Use when you need a conflict-safe set of open issues that agents can work in parallel — a /sge:team-pipeline work pool, "what can we work on at once", "pick the next non-conflicting issue", or blocked vs ready. --fleet spans many repos. Read-only; does not implement.

• build-ready-audit

  Use when gating issues as build-ready before any agent picks them up — clear acceptance criteria, bounded scope, no open questions, dependencies resolved, plus the /sge:governance-trace classification. Writes routing labels; does not implement. Per-spec checks: /sge:sge-preflight.

• clean-root

  Use when a working tree has untracked clutter — stray build logs, leftover report files, byte-identical duplicates of committed content. Removes only provably safe or allowlisted files and reports the rest. Not for worktrees (/sge:tidy-worktrees) or processes (/sge:reap-orphans).

• code-map

  Use when a repo should give GitHub Copilot (or any session) a cheap read of what each component and directory does — generate or refresh docs/sge-code-map.md, point AGENTS.md and .github/copilot-instructions.md at it, and set up the SGE plugin for Copilot. Not for editing architecture.yaml.

• commit

  Use when committing work in a WTP/SGE repo — end of an implementation slice, before opening a PR, or when another skill says "commit via /sge:commit". Runs quality gates and adds SGE traceability trailers. Not for amending history or interactive rebase.

• cost-guard

  Use when checking the session's token use against the active spec's BudgetPolicy — \"are we within token budget?\", before another expensive slice, or when SGE_SPEC_ID is set. Advisory: reports ok/alert/deny and never stops the session. For cost reports use /sge:roi-report.

• decompose-issue

  Use when an issue is too large for one session and should be split into ordered, parallel-safe child issues (an enabler plus vertical slices) — decompose, break down, split or fan out, or /sge:sge-implement sizes it Large. Creates the child issues and plan; does not build.

• deep-dive

  Use when an issue needs investigation and a recorded decision rather than an immediate build — an unclear root cause, blast radius or alternatives to weigh, a suspected duplicate or stale issue, or a /sge:sge-align drift issue to triage. Not for building.

• design-discover

  Use when a repo has no design references to hand to /sge:design-gate — the operator cannot name 2-3 sites they rate. Proposes unverified benchmarks, synthesises patterns, renders variants of one real screen, records the operator's choice as taste-profile.md. Run before design-gate.

• design-gate

  Use when installing the design-quality gate (SPEC-115) in a repo — "install the design gate", "enforce design quality". Extracts a DESIGN.md contract from reference sites (or brand/), scaffolds a /design-system route, and wires .claude/design-review/ and CLAUDE.md.

• dora-setup

  Use when rolling out status monitoring for a new org — stand up an Upptime status repo, GitHub Pages site, custom domain, optional brand skin and a DORA incident signal from scratch. Generic; no org-specific assumptions.

• drift-hillclimb

  Use when a repo's SGE Audit Score (the /sge:sge-align rollup, SM-2) or a drift metric needs raising, not just measuring. Picks the highest-leverage gap, opens one bounded PR, re-measures independently, and repeats until the target or a bound. Advisory, PR-first.

• env-health

  Use when an unattended SGE session's throughput collapses or could (orphaned servers, broken pnpm symlinks, too many sessions, hung install), or when hooks fail with "Hook command failed" (--doctor: which bash, PATH order, versions, fix). Hourly or preflight.

• exit-report

  Canonical exit-report contract for orchestrated SGE skills — one machine-readable JSON shape (schema.json beside this file) that orchestrators consume to decide the next move. Fields — skill, runId, duration, itemsProcessed, outcomes[], stopReason, followUps[]. Emitting skills link here instead of i

• fleet-dispatch

  Use when dispatching build-ready issues across many repos at once — "dispatch across the fleet", "work issues org-wide", "run the pipeline across all our repos" — with at most one lane per repo. Composes /sge:available-issues --fleet and /sge:team-pipeline.

• gh-repo

  Canonical reference for targeting the right GitHub repo when an SGE skill is invoked outside the target repo's checkout — the GH_REPO / cd convention for cross-repo and hub (control-session) dispatch, the startup echo check, and the raw-git pitfall. gh-heavy skills link here instead of restating the

• governance-trace

  Use when classifying an issue against a repo's SGE governance (Vision, Capability Model, Feature Specs) before code is written — matches a spec, modifies one, needs a new spec, needs none, or is out of scope. Dispatched by /sge:sge-implement Phase 0.5 and /sge:deep-dive.

• issue-intake

  Use when analysing issue readiness, acceptance criteria and remaining scope, including unattended batches. Recommends a bounded next step without approving work. --record posts the governance record a PR is checked for; --ready runs the readiness loop and applies the dispatch label.

• issue-loop

  Use when draining a repo's backlog one issue at a time — each build-ready issue taken through implement, review and merge until none are left ("work the backlog one issue at a time"). Serial and queue-bounded; for parallel or time-boxed runs use /sge:team-pipeline.

• lessons-learned

  Use when closing a delivery cycle or on a schedule - one command, no arguments, turns the window's review findings, already-done closures, exit-reports and corrections into classified lessons, one bundled fix issue per recurring class, a lessons log and Cortex memory.

• loops

  Canonical reference for the four loop patterns used across SGE skills — the inner Red/Green/Refactor loop, the wait-for-condition loop, the bounded refinement loop, and the recurring cross-session loop. Other skills link here instead of re-describing a loop; this file is not a user command.

• onboard

  Use when onboarding people to SGE — a guided first-week path for a new developer (--dev) or ceremony-overlay adoption for an existing agile team (--team). Interactive.

• optimisation-loop

  Use when optimising a system whose output is gated by a measurable bottleneck and each unit of effort should yield the maximum, *verified* improvement — e.g. "grow/optimise metric X", "size a data-coverage or backlog effort before doing it", "prove this intervention works". The pattern: find the REA

• pr-fix

  Use when a pull request's CI is red and needs driving to green — failing required checks, lint/test/build failures, or a /sge:pr-monitor CI-failing dispatch. Also fixes a conflicting (dirty) PR, and with --all-prs a whole backlog of red PRs.

• pr-monitor

  Use when several open PRs need shepherding to merge without babysitting — after a batch of PRs lands, when the user asks to monitor/babysit/drive open PRs, when a backlog of red or review-blocked PRs has built up, or for a bounded unattended shepherding pass over a repo's open PRs.

• pr-review

  Use when a pull request needs the SGE merge-gate review — before merging, when a PR is review-blocked (`mergeStateStatus: BLOCKED` or no `pr-reviewed` label), when /sge:pr-monitor routes a PR here, or for a delta re-review after new commits.

• prod-reliability-playbook

  Use when a simple fix drags into an all-day slog, debugging keeps bouncing through production, or a deploy ships plausible-but-wrong output; or before/after an incident. Stack-agnostic playbook — fail loud, shift-left validation, a fast green path, a triage checklist. Advisory.

• qa-audit

  Use when a PR needs runtime QA evidence that the change works in the running app — exercising acceptance criteria or spec Gherkin against a live build, a QA report for a high-risk PR, or (--adversarial) proving a control enforces what it claims. Applies no gate labels.

• reap-orphans

  Use when the box is sluggish or after long sessions — kills orphaned Claude Code processes (claude/node/bash, dev servers with dead parents) and proven-abandoned Copilot CLI orphans, reports hogs; --heavy also kills headless Chromium and stops WSL. Protects live sessions.

• refactor

  Use when the user asks to refactor, restructure, or clean up existing code, or mentions code smells, duplication, a god class or god function, tangled dependencies, dead code, cleanup, or tech debt. Behaviour-preserving restructuring only — not for new features or bug fixes.

• regulatory-trace

  Use when a regulated repo must map specs and capabilities to FCA/UK obligations and ISO 27001/42001/27701 and export that as audit evidence — due-diligence packs, FG26/4 registration, ISO audits — or when /sge:sge-align C12 flags a missing obligation mapping.

• review-daemon

  Operational reference for the SGE review daemon (SPEC-090 Layer 1). Read before operating, configuring, or extending the daemon or its claim-mutex protocol.

• roi-report

  Use when you want a token cost attribution report — "where did our tokens go?", per-spec and per-PR breakdowns, the governed-vs-unattributed gap, or a sprint-end/CI snapshot. Reporting only; for live budget checks use /sge:cost-guard.

• sge-adoption-diagnostic

  Use when a team wants to assess their readiness for SGE adoption before committing — produces a scored readiness assessment and a sequenced rollout plan. Run this before /sge:sge-init for new prospects or teams. Not for teams already running SGE — that is /sge:sge-align.

• sge-ai-inventory

  Use when a financial-services firm must register or review its AI/LLM use cases — FCA/PRA or EU AI Act scrutiny, a DORA third-party register entry, model-risk docs, AI vendor due diligence, "what AI are we running and is it governed?", or a new AI feature with no entry.

• sge-align

  Use when checking a repo or fleet for SGE governance drift — features without specs, specs without Gherkin, ungoverned code, issues out of scope after a change — or for an Audit Score scorecard before a review. --dimension adds agent-security, regulatory or skill-quality scoring.

• sge-dashboard

  Use when a repo should publish, refresh or CI-wire the brand-neutral SGE coherence dashboard (Vision → Capabilities → Features → Specs → Tests → Decisions) generated from docs/sge-dag.json and the SGE markdown. Not for authoring specs or applying WTP's brand.

• sge-implement

  Use when implementing a GitHub issue end-to-end to a merged PR — spec issues (`SPEC-NNN`, legacy `SGD-NNN`) and plain feature/bug/chore issues, including production bugs (`--sentry`) and headless runs (`--unattended`). Use when asked to build, implement, fix or ship an issue.

• sge-init

  Use when onboarding a new product or greenfield repo onto SGE — no Vision, capability model or feature specs yet; "set up SGE", seed governance artefacts, run product intake, or turn a brief into seed artefacts. Not for auditing a seeded repo (/sge:sge-align).

• sge-preflight

  Use when preparing to implement an SGE feature spec — before any branch, plan, or code exists — when asked to preflight or check build-readiness of a SPEC-NNN, or when /sge:sge-implement Phase 1 delegates its entry-criteria checks.

• sge-review

  Use when a finished implementation needs an independent review against its SGE spec — after coding completes and before a PR is opened, when /sge:sge-implement Phase 5 delegates its pre-PR review, or when asked to verify that work satisfies a SPEC-NNN.

• sge-skill-audit

  Use when scoring SGE skill quality against SQ-0–SQ-5 (frontmatter, executability, cost-awareness, scope, UNTRUSTED DATA annotation, tool sequencing) — one skill before merging a SKILL.md change, or a fleet sweep. Delegation target for /sge:sge-align --dimension skill-quality.

• spec-validate

  Use when running a spec's `## Validation` section against a demo fixture to check its business-rule invariants actually hold — after editing that section, at spec graduation (draft → approved → implemented), or when /sge:sge-align C4 flags a missing or unverified section.

• spike

  Use when an architecture bet (a new extraction approach, pipeline, or adapter/integration layer) could be built two ways and the right shape is unknown. Time-boxes a throwaway exploration with a keep/kill checkpoint before the winner goes through the normal SGE pipeline.

• tdd-workflow

  Use when writing the next increment of implementation code — the inner Red/Green/Refactor loop inside any implementation workflow. Triggers on implementing a feature slice, fixing a bug test-first, or any task where production code is about to be written.

• team-pipeline

  Use when running several SGE implementation agents in parallel as a continuous implement-and-review pipeline — "run the pipeline", "work issues in parallel", "batch implement issues". With --duration it is the time-boxed swarm ("swarm the issues for 2 hours") and never overruns.

• tidy-worktrees

  Use when cleaning up git worktrees, branches, stashes or leftover files, after merging PRs or before new work (--force: fast merged-only sweep). Audits everything, builds one delete plan, assesses unmerged work for PR or discard. Destructive only after one confirmed plan with tip SHAs.

• traceability

  Use when installing the SGE Traceability Module (SPEC-054) in a repo — "install the traceability chart/module". Copies the data builder and chart page, wires the docs Pages workflow and opens a draft install PR. Does not build or audit traceability data itself.

• worktrees

  Canonical reference for where SGE skills place git worktrees — the sibling `../<repo>-worktrees/<purpose>-<id>` layout, the in-repo `.worktrees/issue-N` team-pipeline exception, and the deprecated stray layouts. Other skills link here instead of restating placement rules; this file is not a user com

Subagents (3)

• code-reviewer

  Stack-agnostic code reviewer for PR reviews, pre-commit self-review, or auditing existing code. Catches correctness bugs, security issues, performance problems, and maintainability concerns, and validates the change against its stated requirements. Examples: - "Review this PR" - "Check this code bef

• design-reviewer

  Adversarial design QA on the rendered app. Use PROACTIVELY after any UI change, and whenever the design gate demands a review. Reviews either LIVE (Playwright MCP tools) or from a pre-captured static evidence directory (screenshots + measurements.json from scripts/capture-design-evidence.mjs) when n

• security-auditor

  Stack-agnostic application-security reviewer. Use when a change touches authentication, authorization, input handling, secrets/credentials, external integrations, or anything that processes untrusted input. Finds vulnerabilities (injection, XSS/CSRF, auth bypass, insecure deserialization, secret lea

MCP servers (1)

• sge-memory

  runs: node ${CLAUDE_PLUGIN_ROOT}/mcp/sge-cortex/dist/sge-cortex.bundle.mjs

Hooks (3)

• SessionStart

  . "${CLAUDE_PLUGIN_ROOT}/hooks/run-hook.cmd" "${CLAUDE_PLUGIN_ROOT}/hooks/cortex-recall.sh"

• Stop

  . "${CLAUDE_PLUGIN_ROOT}/hooks/run-hook.cmd" "${CLAUDE_PLUGIN_ROOT}/hooks/token-meter.sh"

• SubagentStop

  . "${CLAUDE_PLUGIN_ROOT}/hooks/run-hook.cmd" "${CLAUDE_PLUGIN_ROOT}/hooks/token-meter.sh"