sdlc

Plugin sdlc from marketplace claudecode-patterns in pattern-stack/claudecode-patterns. Commands read from the repo's manifest — 14 days ago.

Skills (17)

• browser-driver

  Drive a real browser against a URL — capture full-page screenshots, inspect computed styles / ARIA / cursor / overflow per selector, and exercise interactions (open menus, hover for tooltips). Wraps `${CLAUDE_PLUGIN_DIR}/scripts/capture-surface.mjs` and `${CLAUDE_PLUGIN_DIR}/scripts/inspect-surface.

• browser

  Browser interaction via the browser-pilot agent's MCP servers — see and drive the user's own browser over CDP, run independent headless checks via Playwright, and audit with chrome-devtools' Lighthouse tool. Auto-invoked when the conversation mentions browser verification, console errors, screenshot

• canvas-authoring

  Meta-knowledge for authoring canvases — the template + instructions pairs that govern artifacts produced by SDLC agents (specs, plans, reports, …). Covers the canvas concept, the canonical knob taxonomy, conversation modes (new / tune / reverse / validate / explain), the response scaffold, and the a

• claude-platform

  Authoritative reference and authoring guide for the Claude platform — both the Claude Code `.claude/` config layer (skills, subagents, output styles, hooks, plugins, settings) and the Anthropic Agent SDK (Managed Agents API: agents, sessions, threads, multiagent, skills). Use when authoring a skill

• critique

  Structured critic discipline for SDLC artifacts — specs, plans, diffs, ADRs. Applies a verdict (PASS / PASS_WITH_NOTES / REVISE / BLOCK) with categorized findings (blockers / notes / nits) against a configurable lens (adherence / quality / logic / scope / mixed). Used by the reviewer agent at Gate 1

• design-audit

  Audit-only design pass. Grader runs once against a built surface; no builder dispatch, no fix loop. Thin wrapper around `/design-loop --mode=audit`. Use when you want a design quality check on a PR or branch without iterating.

• design-loop

  Iterate a UI surface against a design reference (spec / figma / screenshot). Loop is "agent builds → grader screenshots → gives feedback → agent fixes → grader rechecks." Termination configurable. `/design-audit` is a thin wrapper for audit-only runs. Composable into `/develop` for single-issue spec

• driving-mode

  Hands-free mode — switch this project to the Driving output style, so every reply is written to be heard (the user's phone reads it aloud) and long work runs in Herdr tabs. Use when the user says \"driving mode\", \"I'm driving\", \"walking\", \"hands-free\", \"read it to me\"; leave it on \"I'm par

• figma-snapshot

  Snapshot a Figma frame's metadata, tokens, design-context, and screenshot to local files under `.ai-docs/figma/<slug>/` so downstream agents (design-builder, design-grader, spec authors) don't re-call the Figma MCP every iteration. Universal — not tied to design-loop; usable for spec writing, design

• guided-tour

  Run a scripted walkthrough of the app's UI in two modes — narrate (drives the user's real browser with a visible cursor, highlights and captions, for demos and human-watchable walkthroughs) and verify (same steps, no theatre: screenshots, console errors, failed requests, text assertions, report.json

• handoff

  End-of-session ceremony — articulate what's next, write `.ai-docs/handoff.md` (or update the destination it points to), update memory, update CLAUDE.md if structural decisions landed, verify clean working tree. Symmetric counterpart to `/prime`. Use when wrapping up a session, before clearing contex

• prime

  Load session context at cold-start — handoff, tracker snapshot (open-issue board + branch-derived active ticket), recent commits, stack position. Trigger when the user opens a fresh session or asks "where were we", "what's the state", "starting session", "/prime", or otherwise signals they want a st

• project-documentation

  Create and manage durable project documentation — ADRs, specs, RFCs, and architecture docs. Use when the user wants to record an architecture decision, write an ADR or RFC, draft a spec, update a spec after implementation, or discuss where documentation should live.

• sdlc-loop

  Workflow judgment for this project's SDLC loop — when to use /plan vs /design vs /develop vs /orchestrate, what each gate means, how to recover from a halt, how stacks and artifacts flow. Use when starting work on an issue, deciding the next command, interpreting a halt from any phase agent, or reco

• skill-authoring

  Create skills, commands, and agents for THIS project — applying the SDLC architecture (primitives, sdlc.yml, artifact_paths, mission format, topology metadata) on top of the canonical Claude Code platform reference. Use when authoring a new component for this repo.

• stack-management

  Auto-load PR stack context for current branch. Reads stack state and reports position. Use when working on any branch that is part of a PR stack.

• stack

  Manage PR stacks with GitHub's official `gh stack` extension — init, add, view, submit, push, sync, rebase, merge, checkout, link, unstack. Use when the user mentions stacks, stacked PRs, restack, stack submit/merge, branch dependencies, the GitHub stack UI, atomic stack merge, or splitting work int

Slash commands (13)

• /sdlc:answer

  Driving mode — deliver the answer to the work just finished as a message of its own, headline pushed to the phone first, so the user hears the answer without the working notes. Claude queues this on itself through Herdr at the end of a working turn; it is rarely typed.

• /sdlc:browse

  Interact with the user's browser — navigate, click, type, inspect, or anything else

• /sdlc:canvas

  Author / tune / validate / reverse-engineer / explain a canvas (a template + instructions pair under .claude/canvases/<name>/) via the canvas-author agent. Probing schema-aware dialog; never writes without confirmation. v1 ships tune + validate.

• /sdlc:critique

  Gate 1.5 — single-reviewer spec critique. Spawns the reviewer agent with lens=mixed against cited code; writes verdict + findings to the spec's Spec Review phase section. Verdict PASS/PASS_WITH_NOTES clears Gate 1.5; REVISE triggers specifier re-run; BLOCK halts for human arbitration.

• /sdlc:design

  Generate the implementation strategy for a tracker issue and post it. In strict mode (default) sets state:awaiting-strategy-review and halts; in auto mode (per gate:auto label) sets state:strategy-approved directly and completes. Resolution by gate:* label → sdlc.yml.gate1_default.

• /sdlc:develop

  Topology A — work one issue end-to-end with a flat team of split-pane teammates. Composes the team from sdlc.yml develop_team plus the issue's needs:* labels. Accepts a tracker issue key OR an approved spec path.

• /sdlc:doctor

  Diagnose Claude Code / sdlc misconfigurations that silently break the harness — e.g. a WorktreeCreate provider-hook registered as passive telemetry. Read-only; reports findings and fixes.

• /sdlc:orchestrate

  Topology B — parallel batch execution. Spawns one coordinator teammate per issue; each coordinator runs phase agents as headless subagents. Best for AFK throughput on pre-approved issues.

• /sdlc:plan

  Decompose a request into a PR-sized issue plan. Produces a YAML at .ai-docs/plans/<slug>.yaml; iterates against your feedback until you approve. Does NOT touch the tracker.

• /sdlc:review

  Gate 2.5 — paired-lens post-implementation diff review. Spawns two reviewer agents in parallel (lens=adherence + lens=quality), each writing to its own spec phase section. Joins verdicts; the lower wins. Verdict PASS/PASS_WITH_NOTES clears Gate 2.5; REVISE triggers implementer fix.

• /sdlc:setup

  Onboarding flow that scaffolds a fresh project's `.claude/sdlc.yml` + Justfile module + symlink to the plugin's `sdlc.justfile`. Drives an interactive AskUserQuestion dialog (4 questions). Idempotent — re-runs safely with reconfigure prompt.

• /sdlc:sync-issues

  Sync an approved YAML stack plan to the active task-management adapter. Creates the stack's epic parent + leaf issues + sub-issue rollup, idempotently by plan-key marker. Run after /plan ends with "approved". Does not run agents.

• /sdlc:verify

  Run visual QA on a URL — screenshot, console, network, performance

Subagents (12)

• canvas-author

  Probing, schema-aware conversational authoring of canvases (template + instructions pairs at .claude/canvases/<name>/). Use when creating a new canvas, tuning an existing one, reverse-engineering a canvas from an example artifact, validating a canvas, or explaining how knobs affect produced artifact

• claude-platform-drift-check

  Read-only auditor that diffs the project's `claude-platform` skill against canonical Claude Code and Anthropic Agent SDK docs, surfacing renamed / added / removed fields, new lifecycle events, new SDK primitives, and changed precedence rules. Use when the user asks "is this still right", before auth

• coordinator

  Per-issue coordinator for Topology B parallel orchestration. Spawned as a teammate by `/orchestrate`; spawns implementer/validator as **subagents** (Agent tool) — never as further teammates. SKELETON — not yet wired to a command.

• design-builder

  Build or refine a surface against any design reference (spec / figma / screenshot). In phase mode, produce a build matching the reference. In fix mode, apply a numbered findings list from the grader. Replaces v1's design-implementer; works across all three reference types.

• design-grader

  Grade a built surface against a design reference (spec / figma / screenshot). Captures surface evidence via browser-pilot, compares to the reference using the type-specific strategy, posts findings via the image-posting primitive. Returns READY / FIXES / BLOCKED with sub-codes. Replaces v1's design-

• implementer

  Executes an approved spec. Refuses to start unless the issue carries `state:strategy-approved` (Gate 1 enforcement). Creates a branch, writes code per spec, opens a PR. Does not iterate strategy.

• planner

  Decomposes an understood problem into PR-sized issues. Iterates a YAML plan across chat turns based on human feedback (Gate 0 — synchronous, in-chat). Halts only on explicit human approval; does not write to the tracker.

• reviewer

  Runtime role for the critique skill. Reads a target artifact (spec / plan / diff / ADR), reads an `against` reference (spec / quality canvas / cited code), applies a lens (adherence / quality / logic / scope / mixed), and emits a structured verdict (PASS / PASS_WITH_NOTES / REVISE / BLOCK) with cate

• sdlc-author

  Context-isolated SDLC-aware artifact writer for cases where a phase agent isn't right (no workflow to run, no gate to enforce) but `general-purpose` is too broad. Writes spec-like artifacts (ADRs, RFCs, ad-hoc design docs, handoff notes, batch spec drafts) under the SDLC canvas conventions. No Bash,

• specifier

  Turns a planned issue into a per-issue implementation strategy. Posts the strategy as a tracker comment. In strict mode sets `state:awaiting-strategy-review` and halts (Gate 1 — async, tracker-native). In auto mode sets `state:strategy-approved` directly and does not halt. Resolution at runtime by `

• understander

  Produces a detailed understanding artifact for downstream agents AND a condensed chat/comment summary for the human. Stack-co-located by default (`.ai-docs/stacks/<slug>/<topic>.md`); cross-cutting research goes to `.ai-docs/research/<topic>.md`. Read-only on source. Path convention defined in `.cla

• validator

  Verifies an implementation against the configured quality profile and posts a report to the PR. Read-only with respect to source — runs gates, surfaces failures, never fixes them.

Hooks (30)

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh SessionStart

• SessionStart

  mkdir -p ${CLAUDE_PROJECT_DIR}/.claude/.session && bash ${CLAUDE_PLUGIN_ROOT}/primitives/task-management/discover.sh > ${CLAUDE_PROJECT_DIR}/.claude/.session/tracker-context.md 2>/dev/null || true

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/ensure-cc-viewer.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/ensure-statusline.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/ensure-cc-bridge.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/ensure-browser-agent.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/check-cdp.sh

• InstructionsLoaded

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh InstructionsLoaded

• UserPromptSubmit

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh UserPromptSubmit

• UserPromptSubmit

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/check-config.sh

• UserPromptSubmit

  bash ${CLAUDE_PLUGIN_ROOT}/scripts/doctor.sh --hook UserPromptSubmit

• PreToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh PreToolUse

• PreToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/gate-guard.sh

• PreToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/phase-tuning.sh

• PermissionRequest

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh PermissionRequest

• PermissionDenied

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh PermissionDenied

• PostToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh PostToolUse

• PostToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/driving-answer-stop.sh

• PostToolUseFailure

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh PostToolUseFailure

• Notification

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh Notification

• SubagentStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh SubagentStart

• SubagentStop

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh SubagentStop

• TaskCreated

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh TaskCreated

• TaskCompleted

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh TaskCompleted

• Stop

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh Stop

• StopFailure

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh StopFailure

• TeammateIdle

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh TeammateIdle

• ConfigChange

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh ConfigChange

• CwdChanged

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh CwdChanged

• FileChanged

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/emit.sh FileChanged

Similar plugins