███████╗░█████╗░░██████╗░██╗░░░░░███████╗  ░██████╗██╗░░██╗██╗██╗░░░░░██╗░░░░░░██████╗
██╔════╝██╔══██╗██╔════╝░██║░░░░░██╔════╝  ██╔════╝██║░██╔╝██║██║░░░░░██║░░░░░██╔════╝
█████╗░░███████║██║░░██╗░██║░░░░░█████╗░░  ╚█████╗░█████═╝░██║██║░░░░░██║░░░░░╚█████╗░
██╔══╝░░██╔══██║██║░░╚██╗██║░░░░░██╔══╝░░  ░╚═══██╗██╔═██╗░██║██║░░░░░██║░░░░░░╚═══██╗
███████╗██║░░██║╚██████╔╝███████╗███████╗  ██████╔╝██║░╚██╗██║███████╗███████╗██████╔╝
╚══════╝╚═╝░░╚═╝░╚═════╝░╚══════╝╚══════╝  ╚═════╝░╚═╝░░╚═╝╚═╝╚══════╝╚══════╝╚═════╝

15 skills + 13 agents for Claude Code, Codex, Grok Build, and Google Antigravity. Skills give agents domain expertise via slash commands or automatic invocation. Agent files power autonomous reviewers in Claude Code and Grok Build.

Token saver

Eagle Skills includes a compact hook system that saves context window tokens by rewriting verbose Bash commands. 32 regex rules intercept commands like git log, docker ps, and test runners — replacing them with compact equivalents or filtering noisy output. Add your own rules with /eagle-compact-add. Full details →

Install

npm install -g eagle-skills
eagle-skills install

Or run directly without installing:

npx eagle-skills install

Interactive menu lets you pick what to install. Items are symlinked into the selected agent target, so updates are instant via eagle-skills update.

Install for every supported target:

npx eagle-skills install --target all

Supported targets:

1. claude installs skills to ~/.claude/skills and agents to ~/.claude/agents.
2. codex installs skills to ~/.codex/skills.
3. grok installs skills to ~/.grok/skills and agent files to the Claude-compatible agent directory Grok already discovers.
4. antigravity installs skills to ~/.gemini/config/skills and ~/.gemini/antigravity-ide/skills.

Skills

Invoked via slash commands or automatic skill matching in supported agent surfaces.

| Skill | Command | What it does | |-------|---------|-------------| | UX Review | /eagle-ux-review | Expert UX audits grounded in 65+ UX laws, with HTML/Word reports | | Product Diagnostics | /eagle-product-diagnostics | Three-layer data validation: design intent + behavior + outcomes | | Ad Review | /eagle-ad-review | Strategy-first ad creative review across any medium | | Anti-Slop | /eagle-anti-slop | Detect and eliminate AI-generated slop in code, text, and design | | Human Writing | /eagle-human-writing | Rewrite prose so it sounds human, specific, and less AI-generated | | Clean Doc | /eagle-clean-doc | Modern Word documents — Helvetica, monochrome, clean | | Clean Sheet | /eagle-clean-sheet | Modern Excel spreadsheets — consistent styling, frozen headers | | Multi-Stack Scaffolder | /eagle-multi-stack-scaffolder | Research-driven project scaffolding for 13+ stacks | | CLAUDE.md | /eagle-claude-md | Lean project CLAUDE.md + LLM Wiki + Obsidian vault integration | | Bootstrap | /eagle-bootstrap | One-time global setup: behavioral rules, hooks, vault config | | Governance | /eagle-governance | Hook-based engineering governance: policy gates, context handoff, instruction shims | | Feature Flow | /eagle-feature-flow | Structured dev workflow: plan, build, test, review, anti-slop, commit | | Spectral Review | /eagle-spectral-review | Route broad Spectral review requests to the right specialist agents | | Compact Add | /eagle-compact-add | Review and add new rules to the compact token saver | | Tracker | /eagle-tracker | Project trackers + Gantt charts as .xlsx for Excel and Google Sheets | | Maintainable Code | /eagle-maintainable-code | Human-grade code in any language — minimal, direct, test-driven, no over-engineering |

Agents

Dispatched by Claude's Agent tool. Ask Claude to run one by name, or describe what you need — Claude picks the right specialist.

| Agent | Category | What it does | |-------|----------|-------------| | Spectral Triage | orchestrator | Routes broad Spectral review requests to the right specialists | | Spectral Plan | lifecycle | Scopes features, evaluates approaches, produces implementation plans | | Spectral Investigate | lifecycle | Root-cause debugging — traces code paths, tests hypotheses, fixes | | Spectral Ship | lifecycle | Pre-ship pipeline — build/test/lint gates, diff review, PR creation | | Security Audit | review | Vulnerability hunting — injection, auth, data exposure, supply chain | | Architecture Review | review | Structural health — dependencies, coupling, cohesion, patterns | | Performance Review | review | Bottleneck analysis — memory leaks, N+1 queries, bundle bloat | | Code Quality | review | Readability, dead code, duplication, convention violations, test gaps | | UX Code Review | review | Frontend UX — loading states, error handling, form validation | | Accessibility Review | review | WCAG 2.1 AA — semantic HTML, ARIA, keyboard nav, contrast | | API Review | review | Endpoint consistency, contracts, error handling, versioning | | Data Integrity | review | Race conditions, partial writes, missing transactions | | Database Review | review | Schema design, query performance, migrations, indexes |

CLI

eagle-skills install --target all      # Interactive installer for every target
eagle-skills update --target all       # Pull latest and verify symlinks
eagle-skills status --target all       # Show installed items, check for updates
eagle-skills usage --target all        # Show inputs, outputs, and examples
eagle-skills uninstall --target all    # Remove Eagle symlinks
eagle-skills govern apply --target all # Install hook-based project governance

Docs

| Topic | Link | |-------|------| | Installation guide | docs/installation.md | | All agents reference | docs/agents.md | | Hook-based governance | docs/skills/governance.md | | Compact hook system (token saver) | docs/compact-hook.md | | The three-skill pipeline | docs/pipeline.md | | UX Review methodology | docs/ux-review-methodology.md | | Product Diagnostics methodology | docs/product-diagnostics-methodology.md | | Ad Review methodology | docs/ad-review-methodology.md |

Prerequisites

• At least one supported agent surface: Claude Code, Codex, Grok Build, or Google Antigravity
• ffmpeg — required for video input in UX Review and Ad Review (brew install ffmpeg)
• imagemagick — optional, for Ad Review thumbnail generation (brew install imagemagick)

License

MIT