Scott's Agent Skills and Claude Code Setup

Portable agent skills plus a modular Claude Code plugin suite for productive development. The core plugin provides 8 slash commands, 7 specialized AI agents, 29 skills, 5 hooks, and 2 stored templates that produce 3 project artifacts. Specialized sub-plugins add beads epic workflows, browser automation, mutation testing, multi-persona code review, and more.

Quick Install

Install selected skills for Codex, Hermes Agent, Claude Code, or another supported agent:

npx skills add citadelgrad/scott-cc

The interactive installer lets you choose individual skills and target agents. For exact non-interactive commands, paths, verification, and the skills-only boundary, see Install Skills with npx skills.

Install the complete Claude Code plugin, including agents, slash commands, and hooks:

/plugin marketplace add citadelgrad/scott-cc

npx skills add installs portable skills only. It does not install the Claude-specific agents, slash commands, hooks, or sub-plugin runtime wiring.

At a Glance

Type	Count	Names
Commands	8	delegate-first, gha, handoff, security-cheatsheet, thermo-nuclear, google-standard, polyglot-idiom, concurrency-atomicity
Agents	7	api-debugger, backend-architect, deep-research-agent, frontend-architect, refactoring-expert, requirements-analyst, system-architect
Skills	29	init, acceptance-criteria, cli-design, delegate-first, grill-me, adversarial-reviewer, tdd, python-simplifier, typescript-simplifier, go-simplifier, rust-simplifier, swift-simplifier, karpathy-guidelines, property-based-testing, verified-implementation, context7, context-file-optimizer, c4-diagram, writing-about-engineering, writing-skills-excellence, pas-pipeline, reck-factory, thinking-in-systems, emergent-behavior, skillopt-sleep-learned, thermo-nuclear, google-standard, polyglot-idiom, concurrency-atomicity
Hooks	5	terminal-bell (Stop), toon-post-hook (PostToolUse), prefer-modern-tools (PreToolUse), data-layer-guard (PreToolUse), post-compaction (SessionStart after compact/clear)
Templates	3	.pre-commit-config.yaml, CLAUDE.md, AGENTS.md
Sub-plugins	9	beads-epic-builder, browser-automation, research-tools, security-suite, performance-optimization, mutation-testing, review-panel, variant-explorer, triage

---

Commands (8)

Command	Description
/scott-cc:delegate-first	Keep the main conversation clean by forking implementation work to sub-agents.
/scott-cc:gha	Debug failing GitHub Actions runs and audit workflow YAML. Fetches logs via gh CLI, analyzes errors, suggests fixes.
/scott-cc:handoff	Generate a compact session handoff with git state, active work, key files, and concrete next actions.
/scott-cc:security-cheatsheet	Look up OWASP security cheatsheets by topic. Comprehensive security reference for common vulnerabilities and mitigations.
/scott-cc:thermo-nuclear	Zero-mercy structural-simplification review grounded in Cursor's thermo-nuclear-code-quality-review doctrine. Biases toward ambitious rewrites over preserving imperfect-but-working code.
/scott-cc:google-standard	Review against Google's published Standard of Code Review. Favors approving once a change definitely improves code health, even if imperfect, using the Nit:/blocking distinction.
/scott-cc:polyglot-idiom	Per-language idiom review for Java, C++, C#, Ruby, or PHP. Excludes Python, TypeScript, Go, Rust, and Swift, which have dedicated simplifier skills.
/scott-cc:concurrency-atomicity	Concurrency-correctness review for race conditions, TOCTOU, deadlock/lock-ordering, and transactional atomicity, grounded in real CWE reference entries.

---

Agents (7)

Engineering

Agent	Description
backend-architect	Design reliable backend systems with focus on data integrity, security, and fault tolerance.
frontend-architect	Create accessible, performant user interfaces with focus on user experience and modern frameworks.
system-architect	Design scalable system architecture with focus on maintainability and long-term technical decisions.

Analysis

Agent	Description
deep-research-agent	Specialist for comprehensive research with adaptive strategies and intelligent exploration.
requirements-analyst	Transform ambiguous project ideas into concrete specifications through systematic requirements discovery and structured analysis.

Debugging & Quality

Agent	Description
api-debugger	Expert debugger for APIs, Python backends, and JavaScript/TypeScript frontends with integrated browser testing via Playwright MCP.
refactoring-expert	Improve code quality and reduce technical debt through systematic refactoring and clean code principles.

---

Skills (29)

Project Setup

Skill	Description
init	Interactive project scaffolding. Detects what already exists, presents a menu, and sets up only what you select: beads (bd init), CLAUDE.md, AGENTS.md symlink, .envrc, Makefile, and pre-commit hooks.
acceptance-criteria	Generate testable acceptance criteria before creating beads issues or planning implementation work.

Code Quality

Skill	Description
python-simplifier	Simplifies and refines Python code for clarity, consistency, and maintainability. Applies KISS principles, Pythonic patterns, and framework best practices. Use when reviewing or refactoring Python code.
typescript-simplifier	Simplifies and refines TypeScript/JavaScript code for clarity, consistency, and maintainability. Applies KISS principles, modern ES features, and framework best practices. Use when reviewing or refactoring TS/JS code.
go-simplifier	Simplifies Go code using idiomatic error handling, interfaces, concurrency patterns, and standard-library conventions.
rust-simplifier	Simplifies Rust code using idiomatic ownership, error handling, traits, iterators, and concurrency patterns.
swift-simplifier	Simplifies Swift code using idiomatic optionals, protocols, value semantics, concurrency, and API design.
cli-design	Design patterns and conventions for building agent-compatible CLI tools. Covers flag design, output formatting, exit codes, and composability with AI-driven workflows.
delegate-first	Keep the parent conversation clean by forking noisy implementation, validation, and multi-file work to sub-agents.
grill-me	Socratically stress-test requirements, architecture, edge cases, failure modes, and trade-offs before execution; emits a required risk matrix.
adversarial-reviewer	Evidence-based falsification review: validate candidate defects against known controls, separate evidence from severity, and fail closed on unsupported or stale claims.
tdd	Enforce observed Red → Green → Refactor vertical slices and hand surviving behaviors to the mutation-testing sub-plugin.
karpathy-guidelines	Behavioral guidelines to reduce common LLM coding mistakes. Helps avoid overcomplication, make surgical changes, surface assumptions, and define verifiable success criteria.
property-based-testing	Use when implementing serialization/parsing, data transformations, algorithms with mathematical properties, API contracts, or state machines where testing all edge cases is impractical. Describe invariants instead of specific input/output pairs.
verified-implementation	Require authoritative sources for security-critical, financial, protocol, and production-ready implementation decisions.

Documentation & AI Context

Skill	Description
context7	Retrieve up-to-date documentation for any software library or framework via the Context7 API. Use instead of relying on potentially outdated training data.
context-file-optimizer	Audit and rewrite AGENTS.md, CLAUDE.md, and other AI agent context files to be minimal and effective. Based on research-backed principles: verbose context files hurt performance (−2 to −3%) while minimal, tooling-focused files improve it (+4%).
c4-diagram	Generate C4 architecture diagrams using standard Mermaid flowchart syntax. Covers Context, Container, and Component levels with short labels, companion legend tables, and sequence diagrams for runtime behavior. Never uses the broken C4 Mermaid plugin.

Writing

Skill	Description
writing-about-engineering	Use when drafting first-person engineering writing — blog posts, short posts/threads, or postmortems. Produces a conversational-but-rigorous, peer-to-peer voice anchored on the Julia Evans / Simon Willison TIL/blog rhythm.
writing-skills-excellence	Framework for creating, updating, or improving agent skills. Covers structure, frontmatter, when-to-use clauses, and quality principles.
skillopt-sleep-learned	Locally learned agent preferences and procedures maintained by SkillOpt-Sleep.

Automation & Pipelines

Skill	Description
pas-pipeline	Author, validate, and run PAS DOT pipelines across Codex, Claude Code, and Gemini. Detects authenticated subscription CLIs and current models, enforces the conditional-label contract, and covers bounded execution and checkpoint resumption.
reck-factory	Manage the Reck software factory — register repos, run AI tasks in containers, schedule background pipelines, and monitor results via Loki/Grafana.

Systems Thinking

Skill	Description
thinking-in-systems	Apply Donella Meadows' systems thinking framework to map, diagnose, and redesign any system — organizational, technical, ecological, or policy. Covers stocks/flows, feedback loops, system archetypes, leverage points, and concrete intervention recommendations. Use with --design to build a new system, or --focus map/archetypes/leverage to run a partial analysis.
emergent-behavior	Identify system-level behavior arising from component interactions, test whether it is genuinely emergent, seek disconfirming evidence, and recommend measurable interaction-level interventions.

Code Review

Each of these is runnable directly via its own slash command, and can also be invoked automatically by the model or by another orchestrating skill/agent (e.g. review-panel) when its checkpoints are relevant to the diff under review.

Skill	Description
thermo-nuclear	Zero-mercy structural-simplification doctrine grounded in Cursor's thermo-nuclear-code-quality-review skill. Biased against approval by default; blocks on file-size growth, spaghetti-branching, and missed simplification opportunities that other reviewers would let pass.
google-standard	Google's published Standard of Code Review. Biased toward approval once a change definitely improves code health, even if imperfect; separates blocking findings from optional Nit: findings.
polyglot-idiom	Per-language idiom checkpoints for Java, C++, C#, Ruby, and PHP, grounded in the Gemini code-quality research. Excludes Python, TypeScript, Go, Rust, and Swift, which have (or will have) dedicated simplifier skills.
concurrency-atomicity	Four-checkpoint concurrency-correctness review — race conditions, TOCTOU, deadlock/lock-ordering, and transactional atomicity — grounded in fetched CWE reference entries (CWE-362, CWE-367, CWE-833, CWE-667, CWE-662).

---

Hooks (5)

Hook	Event	Description
terminal-bell	Stop	Terminal tab indicator when Claude finishes responding. Sends a BEL character for tab/dock notification, a desktop notification with a brief summary, and supports Ghostty/iTerm2 (OSC 9) and WezTerm (OSC 777).
toon-post-hook	PostToolUse	Encodes large tool responses to TOON format (a compact alternative to JSON) before they enter the context window. Reduces token consumption on verbose MCP and built-in tool outputs. No-op if toon is not installed.
prefer-modern-tools	PreToolUse	Rewrites legacy CLI commands to faster modern equivalents at runtime: grep/egrep → rg, cat → bat --style=plain --paging=never, ls → lsd, ps aux/ps -ef → procs. Safe near-drop-ins only — tools with incompatible flag syntax (fd, dust, choose) are excluded and documented in CLAUDE.md for native use.
data-layer-guard	PreToolUse	Warns and asks for confirmation before an Edit/Write/NotebookEdit touches a data-layer path (migrations, schemas, ORM models — default globs overridable via .data-guard.json) without a same-day DATA-MODEL.md change-log entry. Interactive/planning-time only: silently no-ops in unattended contexts (--dangerously-skip-permissions/mode:agent), deferring to the data-steward review seat for unattended enforcement.
post-compaction	SessionStart (compact, clear)	Restores the active plan and compact recovery context after Claude Code compacts or clears a session.

Hooks only run when this repo is installed as a plugin. The hook wiring lives in hooks/hooks.json, which Claude Code loads and expands ${CLAUDE_PLUGIN_ROOT} from automatically — but only for plugins installed via /plugin marketplace add (or /plugin install). The root .claude/settings.json in this repo ships "hooks": {} on purpose: it is the config Claude Code reads if you just git clone this repo and open it as a plain project, and ${CLAUDE_PLUGIN_ROOT} has no meaning there.

Net effect: a plain git-clone checkout has zero hook enforcement active, including data-layer-guard — the guard that's supposed to stop an Edit/Write from silently touching a migration/schema/ORM file without a DATA-MODEL.md entry. If you clone this repo directly instead of installing it as a plugin, that protection (and the other three hooks) simply never runs; nothing will warn you that it's missing.

If you want the hooks active without installing the plugin, either:

• Install via the marketplace (/plugin marketplace add citadelgrad/scott-cc) so Claude Code wires hooks/hooks.json up for you, or
• Hand-edit your own .claude/settings.json and copy the hook entries from hooks/hooks.json, replacing ${CLAUDE_PLUGIN_ROOT} with the absolute path to your clone (e.g. /Users/you/scott-cc).

---

Templates (3)

Two templates are stored in templates/; the /init skill uses them to produce three project artifacts.

Template	Deployed by	Description
.pre-commit-config.yaml	init skill	Canonical pre-commit hooks: general hygiene (trailing whitespace, file checks), Python (ruff lint+format, ty type check), TypeScript/JS (biome lint+format), security (gitleaks secret scanning).
CLAUDE.md	init skill	Global Claude Code instructions template covering CLI tool preferences, direnv/Makefile/port conventions, uv-only Python, and C4 diagram standards.
AGENTS.md	init skill (symlink)	Global Codex/agent instructions template. Deployed as a symlink pointing to CLAUDE.md so both agents share the same instructions.

---

Sub-plugins (9)

Install from the marketplace:

/plugin marketplace add citadelgrad/scott-cc/<name>

Status reflects real git activity, not a manual label. Derived from each plugin's git log -- plugins/<name>/:

Status	Rule
stable	3+ commits and a commit within the last 60 days
experimental	Fewer than 3 commits, or 3+ commits with the latest activity 61–89 days ago
unmaintained	No history or 90+ days since the latest commit

Plugin	Status	Description
beads-epic-builder	stable	Plan, build, and swarm beads epics — sequential and parallel execution with CE code review.
browser-automation	stable	Browser testing and validation with E2E test generation and UI validation.
research-tools	stable	Learning guides, tech stack research, and technical writing assistance.
security-suite	stable	Security advisory and vulnerability scanning.
performance-optimization	unmaintained	Performance engineering with bottleneck analysis and profiling.
mutation-testing	stable	Comprehensive mutation testing with zombie test detection and automated refactoring.
review-panel	stable	Multi-persona adversarial code and design review panel.
variant-explorer	stable	Parallel blind-builder variant exploration with AC/taste/simplicity judging.
triage	stable	Foundry-resident triage spine: detect → bead → reproduce → fix → gate loop.

---

beads-epic-builder

Plan, build, and swarm beads epics — sequential and parallel execution with CE code review.

Agents (2)

Agent	Description
epic-planner	Plan complete features from initial concept through implementation-ready beads tasks. Orchestrates research, documentation, and task breakdown with approval gates.
feature-builder	Orchestrate complete feature development from epic to deployment. Manages architecture review, implementation, quality gates, and validation using beads for task tracking.

Commands (2)

Command	Description
/build-feature <epic-id>	Build a complete feature from a beads epic with sequential architecture review, implementation, and validation. Supports --resume for interrupted runs.
/epic-swarm <epic-id>	Build all tasks in a beads epic using parallel worker agents in isolated git worktrees, with CE code review after. Options: --max-parallel 3, --no-review, --dry-run.

---

browser-automation

Browser testing and validation with E2E test generation and UI validation.

Agents (2)

Agent	Description
browser-validator	Validate UI implementations using browser automation with Playwright MCP tools for real-time verification and test generation.
e2e-runner	End-to-end testing specialist using browser-use AI automation. Handles authenticated flows with secure credential injection, supports persistent browser profiles for 2FA, and generates Python test scripts.

Skills (2)

Skill	Description
browser-use	Automate browser interactions for web testing, form filling, screenshots, and data extraction.
browser-use-e2e	Generate and run E2E tests using browser-use AI automation. Handles credentials securely via .env.test with domain-prefixed variables.

---

research-tools

Learning guides, tech stack research, and technical writing assistance.

Agents (3)

Agent	Description
learning-guide	Teach programming concepts and explain code with focus on understanding through progressive learning and practical examples.
tech-stack-researcher	Guide technology choices, architecture decisions, and implementation approaches when planning new features or functionality. Invoked proactively during planning discussions before implementation begins.
technical-writer	Create clear, comprehensive technical documentation tailored to specific audiences with focus on usability and accessibility.

Skills (1)

Skill	Description
humanizer	Remove AI writing patterns to make generated text sound more natural and human, based on WikiProject AI Cleanup guidelines.

---

security-suite

Security advisory and vulnerability scanning.

Agents (2)

Agent	Description
security-advisor	Answer security questions, review architecture for vulnerabilities, and provide tailored guidance by searching OWASP cheatsheets.
security-engineer	Identify security vulnerabilities and ensure compliance with security standards and best practices.

Skills (1)

Skill	Description
plan-security-review	Runs a lightweight threat-model checkpoint over a plan/PRD/spec document at the end of planning — trust boundaries, new data flows, authn/authz surface, secrets, third-party deps — producing a CLEAR/TRIGGERED/N/A report and go/no-go. Planning-stage counterpart to review-panel's Security seat, which reviews diffs.

---

performance-optimization

Performance engineering with bottleneck analysis and profiling.

Agents (1)

Agent	Description
performance-engineer	Optimize system performance through measurement-driven analysis and bottleneck elimination.

---

mutation-testing

Comprehensive mutation testing with zombie test detection and automated refactoring.

Agents (5)

Agent	Description
test-quality-reviewer	Orchestrate comprehensive mutation testing workflow for test quality analysis using semantic code mutations and parallel test execution.
test-saboteur	Create semantic code mutations — applies realistic bugs to verify whether test suites catch them.
test-executor	Execute test suites against mutated code and collect detailed results for mutation testing analysis.
test-auditor	Analyze mutation testing results to calculate mutation score and identify zombie tests, redundant tests, and quality issues.
test-refactor-specialist	Consolidate redundant tests and generate improved test suites based on mutation testing analysis.

Skills (1)

Skill	Description
mutation-test	Run comprehensive mutation testing to audit test quality, find zombie tests, and propose refactoring.

---

review-panel

Multi-persona adversarial code and design review panel, vendoring and adapting patterns from compound-engineering, clairvoyance, superpowers, ponytail, and mattpocock/skills.

Commands (1)

Command	Description
/review-panel [base..head | branch | PR] [--mode=agent]	Run the review-panel orchestrator against a diff, PR, or branch — human-interactive by default, or unattended machine output with --mode=agent.

Agents (1)

Agent	Description
clean-room-alternative	Generates a design alternative in isolation, without seeing the first design. Used by design-it-twice when a first design already exists in conversation.

Skills (33)

Panel orchestration & review seats

Skill	Description
review-panel	Orchestrates a full multi-reviewer code-review panel: casts reviewer seats from diff content, runs them concurrently, merges/dedupes findings, fixes, and re-reviews to convergence.
adversarial-reviewer	Red-teams code, PRs, and designs — hunts for bugs, security holes, and hostile/malformed input handling.
design-review	Orchestrates a diagnostic funnel through complexity, structural, interface, and red-flags checks for overall design quality.
domain-modeling	Reviews a domain model against 8 type-driven functional-modeling techniques (illegal-states-unrepresentable, parse-don't-validate, etc.).
code-evolution	Evaluates whether diffs to existing code look designed-in or bolted-on.
red-flags	Scans code against 17 design smells and produces a structured diagnostic report.
strategic-mindset	Assesses whether code reflects strategic or tactical design investment.
ponytail-review	Reviews diffs exclusively for over-engineering — what to delete, replace with stdlib, or simplify.
ponytail-audit	Whole-repo version of ponytail-review — a ranked list of what to delete or simplify.
data-steward	Reviews migration/ORM/schema diffs against DATA-MODEL.md invariants and a 7-item migration-safety checklist.
taste-review	Reviews diffs against TASTE.md's Preferences/Weightings/Anti-preferences, mapping severity from declared strength.
mental-models-adversarial	Pressure-tests the reasoning behind a change — assumptions, incentives, second-order consequences — via fs.blog-inspired mental models.
mental-models-simplifier	Questions whether this is even the right problem/approach, conceptually, via fs.blog-inspired mental models.
mental-models-systems	Evaluates dynamic runtime behavior — feedback loops, bottlenecks, emergence, scale — via fs.blog-inspired mental models.
mental-models-economics	Frames a change as a resource-allocation decision — tech debt, build-vs-buy, vendor lock-in — via fs.blog-inspired mental models.

Design quality lenses

Skill	Description
abstraction-quality	Evaluates whether abstractions are genuinely useful or structurally shallow.
complexity-recognition	Diagnoses what makes code complex and why, via a three-symptom two-root-cause framework.
comments-docs	Reviews comment quality and documentation practices.
deep-modules	Measures module depth — whether the interface is simple relative to its implementation.
error-design	Reviews error-handling strategy against the "define errors out of existence" principle.
general-vs-special	Evaluates whether interfaces are appropriately general-purpose.
information-hiding	Checks for information leakage across module boundaries.
module-boundaries	Evaluates where module boundaries are drawn and whether modules should merge or split.
naming-obviousness	Reviews naming quality and code obviousness.
pull-complexity-down	Checks whether complexity is pushed to callers or absorbed by implementations.
diagnose	Routes a vague symptom to the most relevant design-quality lens via a decision tree.

Grilling / interview

Skill	Description
grill-my-taste	Elicits personal taste via forced-choice interviews, distilling picks into TASTE.md Preferences/Weightings/Anti-preferences.
grill-the-schema	Interviews to build DATA-MODEL.md — entities, invariants, lifecycle, volume/access patterns, Agent boundaries.
grill-with-docs	Challenges a plan against the existing domain model, updating CONTEXT.md/ADRs inline.

Architecture & planning

Skill	Description
adr-skill	Create and maintain Architecture Decision Records with Socratic questioning and an agent-readiness checklist.
design-it-twice	Generates and compares at least two independent design alternatives before committing.
improve-codebase-architecture	Finds deepening/refactoring opportunities informed by CONTEXT.md and ADRs.

Development workflow

Skill	Description
tdd	Test-driven development with the red-green-refactor loop.

---

variant-explorer

Parallel blind-builder variant exploration: spawns N isolated implementations against a spec and acceptance criteria, judges them against AC conformance, TASTE.md, and simplicity, and produces a ranked shortlist.

Commands (1)

Command	Description
/explore-variants [spec] [--n N] [--ac <path>]	Spawn N blind builders in isolated worktrees against a spec + acceptance criteria, then judge the results and produce a ranked shortlist.

Agents (2)

Agent	Description
blind-builder	Builds one complete, independent implementation of a spec inside an isolated worktree, without seeing any sibling variant.
variant-judge	Scores every surviving variant along one judging axis (AC-conformance, taste, or simplicity) and returns a scorecard per variant. Never edits variant code.

Skills (1)

Skill	Description
explore-variants	Orchestrates the panel: gathers input, validates N (refuses N=1, clamps N>6), spawns blind builders, collects results with explicit failure reporting, runs the judge panel, and hands the human a ranked shortlist.

---

triage

Foundry-resident triage spine: detect → bead → reproduce → fix → gate loop turning detector findings into beads, E2E-reproduced fixes, and review-panel --mode=agent gated PRs.

Skills (3)

Skill	Description
triage-spine	Consumes normalized triage items from any registered detector and runs the loop — intake → reproduce → diagnose → fix → gate — filing a bead per item, reproducing the issue E2E before any fix, producing a fix diff via pas-pipeline, and gating it through review-panel --mode=agent.
detectors/lib-upgrades	Scans a project's dependency manifest/lockfile(s) for outdated or CVE-flagged libraries and emits normalized triage items.
detectors/prod-errors	Consumes a log/Sentry-shaped production error source and emits one normalized triage item per distinct error, stack trace carried verbatim in evidence.

Two of five detector slots (lib-upgrades, prod-errors) are implemented in v1; three (system-upgrades, iac-drift, security-advisory-sweeps) are registered but stubbed.

---

Setup Architecture

This plugin is one layer of a three-layer setup system:

Layer	What	How
1 — Machine	Ansible ai-tools role	./bootstrap.sh in macOS-config — clones this repo, installs tools, deploys security configs
2 — Portable skills	This repo (scott-cc)	npx skills add citadelgrad/scott-cc — select skills and Codex, Hermes Agent, Claude Code, or other targets
3 — Claude plugin	This repo (scott-cc)	/plugin marketplace add citadelgrad/scott-cc in Claude Code
4 — Project	/init skill	Run per-project to scaffold CLAUDE.md, AGENTS.md, .envrc, Makefile, pre-commit hooks

Full bootstrap instructions and Ansible configuration: citadelgrad/macOS-config

See docs/setup-architecture.md for context and init sequence diagrams.

---

Code Quality Standards

All code follows these principles (enforced by simplifier skills):

• DRY — Remove duplicate code
• KISS — Straightforward over clever
• Thin Handlers — Business logic in services
• No Hardcoded Values — Use config/env
• No Silent Failures — Fail fast, specific exceptions
• Function Size — ~20 lines max
• No Premature Abstraction — Wait for 3+ patterns

Best For

• Full-stack engineers
• Next.js / React developers
• Python (FastAPI, Django) developers
• TypeScript projects
• Teams using beads for task tracking

Installation

Portable Skills for Codex, Hermes Agent, and Other Agents

npx skills add citadelgrad/scott-cc

For a scripted global install into both Codex and Hermes Agent:

npx skills add citadelgrad/scott-cc \
  --skill acceptance-criteria \
  --skill tdd \
  --agent codex \
  --agent hermes-agent \
  --global \
  --yes

See docs/skills-cli.md for the complete guide.

Complete Claude Code Plugin

/plugin marketplace add citadelgrad/scott-cc

Update Existing Installation

/plugin marketplace update scott-cc

From Local Clone (for development)

git clone https://github.com/citadelgrad/scott-cc.git
/plugin install /path/to/scott-cc

Requirements

• Claude Code CLI
• Works with any project (optimized for Next.js, FastAPI, TypeScript)
• Beads plugin recommended for epic workflows

License

MIT — Use freely in your projects