claude

cct

The SDLC in one self-contained plugin: sdlc-run and every skill, agent and hook it runs, including the UX suite, the product skills, the engineering and data skills run by hand, a project kickoff and the Docker + GitHub Actions blueprint.

0 on GitHubby Calvin Dudekv0.9.4updated 4 days ago
/

Plugin cct from marketplace cct in cdudek/calvins-claude-tools. Commands read from the repo's manifest — 3 days ago.

This plugin runs code on your machine — its hooks fire automatically on events like tool use. The exact commands are listed below; read them before installing.

What it installs

Skills (77)

  • arch-decision-log

    Records an accepted technical decision as an ADR: context, decision, consequences, rejected alternatives, status and supersession. Writes the ADR file; does not decide. Use when the user says 'record an ADR', 'log this architecture decision', 'write the decision record', 'supersede that ADR'. Skip f

  • arch-plan-write

    Authors a technical proposal or plan: approach with rejected options, API contract, data model, structural-risk questions, migration, order of work and risks. Output is a plan.md in a change folder or an RFC. Use when the user says 'write a technical plan', 'draft a technical RFC', 'plan the API con

  • aws-manage

    Runs the AWS CLI with your credentials (env, AWS_PROFILE or SSO, ~/.aws, or keys in ~/.zshrc) without ever reading or printing them. Use when the user mentions AWS, S3, EC2, Lambda, IAM, CloudFront, Route 53, RDS, DynamoDB, SES, SQS, SNS, CloudWatch, Bedrock, an AWS bill or cost, or asks to list, in

  • browser-automate

    Browser automation CLI (agent-browser) for ad-hoc interaction with websites — navigate, fill forms, click, screenshot, extract data, scrape, verify UI changes. Use when the user says 'open this URL', 'scrape', 'screenshot the page', 'fill out this form', 'check what's on', 'visit', 'navigate to', or

  • cloudflare-manage

    Runs the Cloudflare API and wrangler with the token from your shell (CLOUDFLARE_API_TOKEN), without ever printing it. Zones and nameservers, DNS records, SSL modes, zone settings, Origin, Transform and cache rules, Workers and wrangler deploys, Pages, R2, and the CF-IPCountry country header. Use whe

  • codex-run

    Run the OpenAI Codex CLI headlessly inside Claude Code against a prompt and a target, with a chosen model and reasoning level. A general Codex-exec wrapper — use it to get a non-Claude model to review, refactor, explain, or propose edits. Use when the user asks to "run codex", "codex review", "codex

  • data-classification-metrics

    Reference for reading a classifier's results: the confusion matrix, accuracy, precision, recall, F1, specificity, ROC AUC, which metric fits which cost of error, and how to explain each to a non-technical reader. Grounds other work (reviewing a model, a report or a moderation rule) and writes nothin

  • data-cohorts-analyze

    Analyzes supplied cohort data (CSV, Excel or JSON) and returns insight, writing nothing back: retention curves and heatmaps, feature adoption trends, drop-off points, segment-level differences and anomalies, plus the qualitative research and experiments worth running next. Use when the user says 'co

  • data-network-analyze

    Graph and network analysis — similarity matrices, graph construction, community detection, centrality, and layman explanations of network metrics. Use when the user is asking about relationships between entities. Triggers on "network analysis", "graph analysis", "similarity matrix", "community detec

  • db-query-optimize

    Optimizes PostgreSQL queries and database performance on an existing schema using measured execution: EXPLAIN ANALYZE, indexing, query rewrites and Supabase Postgres practices. Use when the user says 'slow Postgres query', 'optimize this SQL query', 'explain analyze', 'add an index for this query',

  • design-frontend-write

    Implements a supplied visual specification (brief, DESIGN.md, visual contract, design plan) as distinctive, production-grade frontend code, and with an alternatives parameter produces several meaningfully different coded versions side by side. Use when someone says 'implement this visual design', 'b

  • design-mockups-create

    Creates high-resolution UI mockups as generated images from supplied specifications and visual contracts: the OpenAI Codex CLI's image generation, prompted with the design system and fed reference screenshots of the product's own screens. Explores three directions, or draws every screen and state of

  • design-surface-review

    Reviews an authored visual plan or an already-changed UI surface before or at build time, rating it 0 to 10 per design dimension, calibrated against the repo's design system (DESIGN.md or token files), and listing the design issues to fix. Use when the user says 'review this UI plan', 'rate the desi

  • design-system-write

    Authors a product's visual system as a DESIGN.md: typography, color, spacing, radius, layout grid and breakpoint tokens, plus component conventions, for greenfield products. Use when someone says 'set up a design system', 'create DESIGN.md', 'choose fonts and a color palette', 'define the layout gri

  • docs-audit

    Audits repository documentation against current source code and tooling in two modes: diff (changed files only, for PR prep) and full (every doc, for periodic health checks). Use when the user says 'audit the docs', 'full doc audit', 'are the docs up to date', 'stale docs', 'verify the README'. Skip

  • docs-c4-create

    Creates C4 architecture diagrams for a system: context, container and component levels, from the code and configuration. Use when the user says 'C4 model', 'C4 diagram', 'system context diagram', 'container diagram', 'component diagram'. Skip for: ER and data model diagrams or structural-risk planni

  • docs-changelog-write

    Authors release notes and CHANGELOG.md entries from supplied change history (merged PRs, commits, tickets): grouped by Keep a Changelog categories, user-facing wording, PR links, breaking changes called out. Use when the user says 'write the changelog', 'draft release notes', 'write the CHANGELOG en

  • docs-transcript-fix

    Corrects speech-to-text transcription errors using dictionary rules then AI analysis, and keeps a personal correction database that grows with each fix. Use when the user says 'fix this transcript', 'clean up meeting notes', 'fix transcription errors', 'ASR cleanup'. Skip for: auditing documentation

  • git-commits-squash

    Squashes commits into one conventional commit. Branch mode analyzes a feature branch, composes the commit message, discusses concerns and squashes locally with a soft reset. PR mode composes the message from the PR's context and delegates the server-side squash-merge to the scripts of cct:github-man

  • git-pr-run

    Coordinates opening a GitHub pull request from local changes: branch creation per repo conventions, staging without secrets, lint and syntax checks, documentation freshness check (delegated to cct:docs-audit), commit, draft PR publication and CI monitoring. It always opens a draft; only the author m

  • git-release-run

    Coordinates a release: picks the semantic version bump, delegates the changelog and release-note text to cct:docs-changelog-write, updates CHANGELOG.md, tags, pushes the tag and publishes the GitHub release (draft first if asked). Use when the user says 'cut a release', 'tag the release', 'bump the

  • github-manage

    Persists and inspects supplied GitHub resources with the gh CLI: lists PRs and issues, reads CI and Actions status, merges, labels, edits repo settings, runs raw API queries. Delegates domain authoring: changelog text to cct:docs-changelog-write. Use when the user says 'list PRs', 'check CI status',

  • linear-manage

    Persists supplied Linear records through the Linear MCP: creates and updates issues, projects, initiatives, documents, status updates and milestones from text that is already written. Delegates the authored content: work items to cct:product-workitem-write, technical plans and RFCs to cct:arch-plan-

  • posthog-manage

    Runs PostHog from the shell with your personal API key, never printing it - feature flags and rollouts, experiments and their results, dashboards, insights, trend, funnel and retention reports, HogQL, session recordings and console logs, persons, cohorts, surveys, annotations, error tracking, event

  • product-audience-write

    Writes the canonical audience definition of a product: the ideal customer profile (firmographics, behaviours, pain points, disqualifiers, the ideal-of-the-ideal), primary versus secondary audiences with a rule for who wins when they conflict, and named personas with role, goals, frustrations and cur

  • product-discovery-run

    Coordinates a product discovery cycle over 2 to 4 weeks: frame the problem, plan and conduct customer interviews, synthesise insights, design and run experiments, then decide GO, PIVOT or KILL. Delegates problem framing to cct:product-problem-write, journeys to cct:ux-journey-create and ranking of p

  • product-epic-analyze

    Analyzes one oversized epic and returns its vertical slice boundaries using the nine story-splitting patterns (workflow steps, business rules, data variations, interface variations, major effort, simple/complex, operations, defer performance, break out a spike): which slices, in what order, why each

  • product-playbook

    Index of product-management methods and their shared guidance: OKRs, product strategy canvas, go-to-market, discovery interviews and scripts, opportunity solution trees, sentiment analysis, feature-investment cases, unit economics, finance metrics, monetisation, pricing, competitor analysis, battlec

  • product-prd-write

    Writes and maintains the stakeholder PRD that people read: problem, segment, success metric and non-goals on page one, no solution until the problem is agreed, then solution, EARS requirements, decisions log, cross-functional launch readiness and a post-launch impact section. Links the canonical wor

  • product-priorities-analyze

    Returns comparative priorities for roadmap and strategy decisions, writing nothing back: recommends the prioritization framework that fits the product stage, team, decision need and data (ICE, RICE, value/effort, Kano, opportunity scoring) with implementation steps and reassessment triggers, or tria

  • product-problem-write

    Writes the problem statement that briefs and PRDs reuse, from the user's side: persona, desired outcome, barrier, root cause, felt impact, constraints, then one sentence. Two methods: frame a problem from existing research, or reframe a stuck one with the MITRE Problem Framing Canvas (look inward at

  • product-roadmap-write

    Writes the strategic roadmap: orders supplied epics and initiatives into a quarterly or release sequence with priorities taken as given, stakeholder alignment, dependencies and milestones. Use when the user says 'plan a roadmap', 'build a quarterly roadmap', 'sequence our initiatives', 'roadmap for

  • product-storymap-create

    Creates a Jeff Patton user story map from supplied stories: narrative flow of activities and steps on top, story details below, a walking skeleton, and release slices that each deliver an end-to-end outcome. Runs as a facilitated workshop. Use when the user says 'story mapping', 'user story map', 'w

  • product-strategy-run

    Coordinates a product strategy cycle over 2 to 4 weeks, from positioning and customer context through problem framing, solution options and prioritisation to a stakeholder readout and a roadmap handoff. Delegates problem framing to cct:product-problem-write, audience to cct:product-audience-write, p

  • product-workitem-write

    Writes one canonical work item (feature or user story, discovery or spike) from rough product input, and renders it in the requested `format`: `business` (technology-agnostic WHAT, ready for a GitHub issue) or `agent` (self-contained, spec-driven for a coding agent: EARS requirements, Gherkin accept

  • project-kickoff-run

    Sets up a new product before any app code exists - the context repo (product, ICP, personas, jobs to be done, market, roadmap, glossary), the brand guide, the design system and DESIGN.md, the UX docs (interaction guidelines, information architecture and sitemap, owner's words) and the repos from the

  • project-plan-write

    Authors the execution plan for a project or initiative: goal, non-goal, scope, success metrics, milestones, owners and dependencies, definition of done. Text only; creating it in Linear is separate. Use when the user says 'write a project plan', 'plan the milestones', 'draft the initiative', 'assign

  • project-report-write

    Authors a status report or an investigation report strictly from evidence the user supplies: progress, metrics, risks, decisions needed, health; or questions, findings, confidence and recommendations. Use when the user says 'write a status update', 'write the project report', 'write up the spike fin

  • project-repository-run

    Brings a new or existing repo to the blueprint baseline - declares the workflow plugins in .claude/settings.json (cct, optionally pinned), runs the sdlc adopt step, points to the anti-slop install for TypeScript, adds the Decision4 handoff contract and a starter boundary-map.yaml, then checks it all

  • review-code-run

    Coordinates a code review as independent evidence plus finding validation: PR modes (pr, pr-deep, pr-max), local changes (changes, changes-max) or specific files (files, files-max), with a second independent model (OpenAI Codex CLI) when it is installed. It delegates the review passes to the agents

  • review-container-audit

    Audits a Dockerfile, its .dockerignore and build workflow for container hardening: non-root user, multi-stage builds, digest-pinned bases, secrets baked into layers, build-arg credential leaks. Use when the user says 'audit this Dockerfile', 'harden this image', 'docker security check', 'is this ima

  • review-dependency-run

    Runs the review of a dependency update (Dependabot or manual bump) as one workflow: researches changelogs and breaking changes, cross-checks how the repo uses the package, fixes failures the bump causes, scores the risk and posts findings. Auto-detects the ecosystem (Ruby, Node, Python). Use when so

  • review-feedback-analyze

    Investigates code-review feedback and returns, per item, a diagnosis and a disposition without writing anything back: the intended behavior, the root cause, whether it is a single instance or a pattern, a category tag and an action (fix, add test, investigate, defer, discuss, skip). The feedback can

  • review-feedback-run

    Coordinates handling of code feedback from a GitHub PR (worktree, CI failures, merge base, push, per-thread resolve), from findings in the conversation, from free text or from a fix set: it delegates diagnosis and disposition to cct:review-feedback-analyze, then drives test-first remediation (a fail

  • review-fix-run

    Coordinates an automated review, validate, fix, commit loop on local work with fresh subagents per iteration (modes: changes, changes-max, files, files-max) until no findings remain or the iteration limit is hit. It delegates the review to cct:review-code-run (agent cct:review-code-runner), triage o

  • review-security-run

    Runs a security-focused differential review of a code change: auth, crypto, external calls, blast radius and adversarial modeling, with methodology, vulnerability patterns and a structured report. It is a complete review workflow of its own, not a pass of the general code review. Use when someone sa

  • role-code-analysis

    Shared code analysis patterns — intent parsing, root cause analysis, pattern sweeps, categorization, validation, and synthesis. Not a slash command — auto-installed as a dependency.

  • role-review-vocabulary

    Shared vocabulary for code review — severity levels, categories, confidence thresholds, finding format, action verbs, and triage rules. Not a slash command — auto-installed as a dependency.

  • sdlc-run

    Runs an AI-native SDLC in any repo: the agent drafts intent, UX plan, facts and tech plan, the engineer confirms twice, then one /goal builds every feature as commits on a milestone branch and closes the milestone with the heavy checks once and one PR (one go, the default; per-feature PRs remain as

  • test-e2e-coverage-run

    Coordinates E2E test coverage from planning through generation, execution and healing, taking its source from a journey inventory, a Linear ticket or a live surface: plans scenarios, has them generated as live-verified Playwright specs proven to fail when the feature breaks, runs them, and repairs s

  • test-e2e-policy

    The rulebook for end-to-end tests, the automated checks that click through the product like a customer would: which customer journeys deserve one, how a test finds buttons and text so wording or design changes do not break it, how it waits without flaking, which states (empty, error, refused, expire

  • test-e2e-setup

    Configures the named E2E harness (Playwright) and its guards in a repo: installs the lint preset, the copy-catalogue rule, the t() copy fixture, the token helper, the persona and journey templates and the assertion guard, and covers Playwright install, config, auth setup, CI and debugging. It detect

  • test-flake-fix

    Fixes one flaky test, a test that passes and fails at random: reproduces it and calibrates the failure rate, bisects order-dependent cases, applies the minimal fix and proves it with stash-based confirmation. Works from a spec path or a Linear ticket and a seed, for RSpec, Minitest, Pytest, Jest, Vi

  • test-journeys-write

    Writes the canonical journey coverage inventory (e2e/journeys.md): the list of claims the E2E suite makes, grouped by audience and persona, with a recommended score, tier, owner, lifecycle and screen states per journey, every existing spec mapped to a journey or marked a harness, and the uncovered t

  • test-suite-run

    Executes a supplied test suite of any layer (unit, integration or E2E), or a path inside it, with the project's own runner or Playwright, and returns the results: pass, fail, skip and flaky counts, each failing test with its error, and the failure context from the run output. It runs and reports; it

  • tracking-apply

    Adds the analytics one change needs through the project's central tracking library - from a feature, plan or success measure decides each event (which tool, new event or property), adds registry entries, wires call sites through the typed helper, adds tests, regenerates the tracking plan and defines

  • tracking-audit

    Read-only check of a repo or diff against the tracking conventions - SDK calls outside the central library, events not passed as registry keys, bad names, missing registry fields, private data, events before consent, GA4 name and limit breaches, renames that double-count, dead or unregistered events

  • tracking-library-run

    Builds, repairs or migrates a project's central analytics tracking library - typed event registry, one track helper, consent-gated adapters, privacy guard, lint rule, tests and a generated tracking plan for PMs - for web, mobile or server-only apps, and moves old analytics calls onto it with a renam

  • ux-accessibility-audit

    Audits web applications for WCAG 2.2 conformance (Level A and AA, AAA listed) and owns every WCAG verdict, including the overlapping reflow, text-resize and target-size criteria. Does not cover EU Accessibility Act or EN 301 549 requirements beyond WCAG. Use when the user says 'accessibility audit',

  • ux-ai-heuristics

    Reference criteria for judging AI-driven interfaces: Nielsen's ten heuristics adapted to probabilistic, generative and conversational systems, plus AI-specific heuristics (calibrated trust, graceful degradation, feedback effectiveness, transparency of limitations, appropriate autonomy). Grounds cct:

  • ux-assessment-run

    Coordinates specialist UX assessments and merges their verdicts into one findings document, at one of two scopes. scope=app audits the whole running app (every screen and state, every lens, three independent evaluators, an uncapped findings document, sitemap, glossary proposal and proposed guideline

  • ux-brief-write

    Writes a change-specific UX brief: problem space, audience, goals, constraints (scope, technical, brand, timeline, legal), context and inputs, deliverables, and measurable success criteria, focused on the problem rather than a predetermined solution. Use when someone says 'write a design brief', 'wr

  • ux-build-validate

    Returns a built-versus-approved-spec verdict (pass or fail) for one change, from state-by-state visual evidence of the running build compared with the spec and prototype at 100 and 150 percent zoom, with up to three fix rounds. Writes qa.md into the spec folder. Use when the user says 'compare the b

  • ux-copy-write

    Writes and reworks localized interface wording and owns the product's terminology: button labels, CTAs, error, empty-state, toast and confirmation copy, onboarding, tooltips and placeholders, in the repo's voice with locale parity. Use when the user says 'write the microcopy', 'better error message'

  • ux-decision-log

    Records the owner's interaction decisions: turns UX findings and proposed interaction rules into decisions the owner can make without reading anything else (shared rules first, five related decisions at a time, each accepted, swapped, deferred or answered in the owner's words), then writes one decis

  • ux-design-run
  • ux-guidelines-write
  • ux-jobs-write
  • ux-journey-create
  • ux-responsive-audit
  • ux-rules-validate
  • ux-run
  • ux-setup
  • ux-sitemap-create
  • ux-spec-validate
  • ux-states-design
  • ux-usability-audit

Subagents (15)

  • review-code-reviewer

    Reviews a code surface (diff or files) on ONE named axis and returns structured findings. Read-only leaf used by every review-code-run mode — deep passes a full-spectrum axis, max passes one axis per reviewer.

  • review-code-runner

    Runs one review-code-run command (changes / changes-max / files / files-max) end-to-end and reports the $STATE_DIR it produced. Thin command-runner used by review-fix-run.

  • review-criteria-auditor

    Grades a change against the acceptance criteria of its ticket — one verdict per criterion, with the file:line that satisfies it. Read-only leaf used by review-code-run when INPUT.md carried acceptance criteria.

  • review-dimensions-selector

    Selects which review dimensions are worth running against a specific code surface — emits a flat, scored list of axes (core + discovered). Read-only leaf used by review-code-run max modes.

  • review-feedback-runner

    Applies a review-code-run fix set by running review-feedback-run's autonomous fix-set mode. Thin executor used by review-fix-run.

  • review-finding-validator

    Validates a single code review finding by checking evidence and tracing the claimed defect. Returns structured verdict (confirmed/unverifiable/rejected).

  • review-fix-implementer

    Applies the minimal production-code fix that makes a failing test pass, fixing every affected location for PATTERN findings. Phase GREEN leaf used by review-feedback-run's execute-phases.

  • review-fixset-triager

    Triages a review-code-run run's validated findings into a fix-eligible set for the fix loop — buckets by disposition, re-validates unverifiable findings at higher effort, and writes fix-set.json + discuss.json. Inner subagent used by review-fix-run.

  • sdlc-feature-builder

    Builds one feature or one fix part of an sdlc-run change from the plan it is handed: tests first from the facts, code in the plan's folders, a commit, then a short report. With Mode: close it runs a milestone's local pipeline instead. Used by sdlc-run's build stage.

  • test-behavior-verifier

    Verifies one sdlc-run M/L change by exercising changed behaviour and two neighbouring flows, then reports evidence. Report-only subagent; never edits.

  • test-behavior-writer

    Writes one test encoding a stated expected behavior, runs it, and reports pass/fail — without touching production code. Phase RED leaf used by review-feedback-run's execute-phases.

  • test-e2e-generator

    Turns an E2E plan (or a surface) into a runnable, live-verified Playwright spec, driving the agent-browser CLI to confirm selectors (not Playwright MCP). Emits plain deterministic @playwright/test and iterates until green. Used by /cct:test-e2e-coverage-run.

  • test-e2e-healer

    Runs the Playwright E2E suite and repairs failing specs against the live UI, one at a time, without weakening assertions. Re-derives selectors with the agent-browser CLI when needed (not Playwright MCP). Used by /cct:test-e2e-coverage-run.

  • test-e2e-planner

    Explores an app surface with the agent-browser CLI and produces a deterministic E2E test plan (markdown) for the project's Playwright suite. Driver is agent-browser, NOT Playwright MCP. Used by /cct:test-e2e-coverage-run.

  • ux-usability-reviewer

    Critiques and improves UX/UI in one of three modes — `review`, `consult`, or `copilot`. Review mode performs a read-only multi-angle UX critique of a diff or files and returns canonical findings (visual hierarchy, contrast, accessibility, friction, motion, microcopy); use it in parallel review pipel

Hooks (5)

  • PostToolUse

    "${CLAUDE_PLUGIN_ROOT}/hooks/post-edit-lint.sh"

  • WorktreeCreate

    "${CLAUDE_PLUGIN_ROOT}/hooks/worktree-create.sh"

  • WorktreeRemove

    "${CLAUDE_PLUGIN_ROOT}/hooks/worktree-remove.sh"

  • SessionStart

    python3 "${CLAUDE_PLUGIN_ROOT}/hooks/project-drift-check.py"

  • PreToolUse

    python3 "${CLAUDE_PLUGIN_ROOT}/hooks/local-gate.py"

View source on GitHub

Similar plugins

caveman

claude

Talk like caveman. Cut filler. Keep technical details explicit.

22 skills · 1 command · 3 agents · hooks

JuliusBrussee/caveman 119,462

claude-mem

claudecodex

Persistent memory for Claude Code: the next session starts where the last one left off.

29 skills · 1 MCP server · hooks

thedotmack/claude-mem 116,618

Taste Skill

claudecodex

tasteskill 2.5.1: professional company websites at the figma.com and x.ai level, with one section per screen, real photographs, real product UI, effects kept for tech and gaming pages, and a real-browser checker and finish gate.

1 skill

Leonxlnx/taste-skill 114,391

career-ops

claudecodex

Open-source AI job search that runs locally in your AI coding CLI: scan job portals, evaluate listings into a structured A-H report with a 1-5 score, tailor your CV, track applications

5 skills · hooks

career-ops-hq/career-ops 82,253

mem0

claudecodex

Cross-session memory and token savings for coding agents.

6 skills · 1 agent · 1 MCP server · hooks

mem0ai/mem0 68,893

pake

claudecodex

Package any website or local web build into a lightweight desktop app with Pake CLI.

1 skill

tw93/Pake 63,463

Maintain cct?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex pluginslisted