AugmentClaude

Agentic Code Orchestrator

Build, analyze, and deliver code projects with unified codebase manipulation and data analysis.

Installation

  1. Make sure Claude is on your device and in your terminal.

    Skills load from ~/.claude/skills/ when Claude Code starts up β€” so you need it on your machine first. If you don't have it yet, install it once with the command below, then run claude in any terminal to verify.

    One-time setup
    npm i -g @anthropic-ai/claude-code

    Already have it? Skip ahead.

  2. Paste into Claude Code or into your terminal.

    This copies the whole skill folder into ~/.claude/skills/agentic-code-orchestrator-winstonkoh87/ β€” the SKILL.md plus any scripts, reference docs, or templates the skill ships with. Works for every skill.

    What this command does
    1. Makes a new, empty temporary folder.
    2. Downloads the skill's folder, examples/skills/coding/agentic-code-orchestrator, from github.com/winstonkoh87/Athena-Public (branch main), the GitHub repository it's listed from. You get the version that is there now, not a copy we keep.
    3. Copies that folder to ~/.claude/skills/agentic-code-orchestrator-winstonkoh87/.
    4. Deletes the temporary folder.

    Installing runs none of the skill's files; it only downloads and copies them. When you later use the skill, Claude follows its SKILL.md and may run scripts that come with it. Read them first.

    How we check skills, and what we don't Β· Report this skill

    Faster alternative (instruction-only skills)

    Skips the clone and grabs only the SKILL.md file. Don't use this if the skill ships Python scripts, reference markdowns, or asset templates β€” they won't be downloaded and the skill will fail when it tries to load them.

    Quick install (SKILL.md only)
    Sign up to copy
  3. Restart Claude Code.

    Quit and reopen Claude Code (or any other agent that loads from ~/.claude/skills/). New skills are picked up on startup.

  4. Just ask Claude.

    Skills auto-activate when your request matches the skill's description β€” no slash command needed. Trigger phrases live in the skill's own frontmatter; you can read them in the β€œWhat this skill does” section above.

Prefer to read the source first? Open on GitHub.

When Claude uses it

Unified codebase manipulation, AI deployment, data analysis, and academic delivery engine. Absorbs 6 coding protocols + data-analysis + academic-delivery + spec-driven-dev.

What this skill does

Agentic Code Orchestrator β€” Build Γ— Analyze Γ— Deliver

Compiled: 2026-05-11 (retroactive synthesis of all engineering/academic sessions) Problem Class: All code generation, data analysis, dashboard building, academic delivery, and technical project execution. Axiom: "The winner is not who thinks deepest on the first try β€” it's who can iterate fastest at the lowest cost per loop."

When to Use

Invoke whenever the user mentions:

  • Building/fixing a website, dashboard, or web app
  • Data analysis (CSV, JSON, Parquet, large datasets)
  • Academic assignments (essays, capstones, SUSS coursework)
  • Code refactoring or architecture decisions
  • Deploying to Supabase, Vercel, GitHub Pages
  • "Analyze this data" / "Build me a [thing]" / "Fix this bug"

Solution Architecture

Module 1: The RETO Engine Selector (COD-415 + MP-2)

Before writing ANY code, classify the project:

Is failure reversible?   β†’ Efficient Engine (Vibe Engineering: ship at 70%)
Is failure irreversible? β†’ Robust Engine (Nuclear Plant: test everything)
Project TypeEngineTest CoverageShip Threshold
Portfolio/websiteEfficientVisual QA only70%
Client dashboardEfficient→RobustVisual QA + data validation85%
Financial calculationsRobustUnit tests + manual verification99%
Academic submissionRobustPlagiarism check + format audit95%
Quick prototype/MVPEfficient"Does it work?"60%

Module 2: Spec-Driven Development (COD-107)

NEVER build without a spec. The spec is the contract.

Phase 1: Interrogation (The /brief)
  β†’ What does the user ACTUALLY want?
  β†’ What are the constraints?
  β†’ What does "done" look like?

Phase 2: design.md Generation
  β†’ Architecture diagram
  β†’ Component breakdown
  β†’ Data flow
  β†’ Acceptance criteria

Phase 3: User Approval
  β†’ Review the spec
  β†’ Confirm scope
  β†’ THEN and ONLY THEN β†’ build

Phase 4: Execution
  β†’ Build to spec, not to vibes
  β†’ Checkpoint every major component

Module 3: The De-Sloppify Protocol (ECC Steal)

After generating code, ALWAYS run this quality pass:

  1. Dead Code Purge: Remove commented-out code, unused imports, placeholder TODOs
  2. Console.log Sweep: Remove all debug logging from production code
  3. Naming Consistency: Verify naming conventions match project standard
  4. Error Handling: Ensure every async operation has error handling
  5. Type Safety: If TypeScript, no any types unless explicitly justified

Module 4: Data Analysis Pipeline (DuckDB-Powered)

For large data dumps (CSV, Parquet, JSON):

Phase 1: Ingest
  β†’ Identify file format + encoding
  β†’ Load with DuckDB (NOT Pandas for large files)
  β†’ Profile: row count, columns, types, nulls, distribution

Phase 2: Profile
  β†’ Summary statistics per column
  β†’ Outlier detection
  β†’ Cardinality analysis
  β†’ Missing data assessment

Phase 3: Query
  β†’ User-directed analysis
  β†’ SQL-based queries via DuckDB
  β†’ Visualization where appropriate

Phase 4: File Insights
  β†’ Key findings summary
  β†’ Actionable recommendations
  β†’ Export results

Rule: For files >100MB, ALWAYS use DuckDB. Pandas will crash.

Module 5: Academic Delivery Pipeline

For SUSS assignments, essays, capstones:

Step 1: Intake β€” Parse assignment brief, identify marking rubric
Step 2: Research β€” NotebookLM arbitrage for source material
Step 3: Outline β€” Structure mapped to rubric weightings
Step 4: Draft β€” Write with burstiness and perplexity variation
Step 5: Red-Team β€” Invoke red-team-review on key arguments
Step 6: Humanize β€” Run academic-humanizer if AI detection risk
Step 7: Format β€” APA/Harvard citation formatting
Step 8: Deliver β€” Final audit against rubric

The Bionic Academic Advantage (CS-467):

  • AI drafts at 80%, human polishes to 100%
  • Research Arbitrage: NotebookLM handles volume, Athena handles synthesis
  • SPSS/R/Python for statistical analysis (statistical-analysis skill)

Module 6: Dashboard/Website Architecture

For financial/trading dashboards:

PrincipleRule
Decimal Standard4 decimal places for all statistical outputs (GTO compliance)
Render StabilityExtract primitive values for useEffect deps, never use object refs
Visual HierarchyStatus indicators (green/amber/red) for institutional readability
ResponsiveMobile-first, then desktop adaptation
PerformanceLazy load heavy components, debounce real-time updates

For portfolio/marketing websites (CS-437 UI/UX Pro Max):

PrincipleRule
Above-the-foldHero β†’ problem statement β†’ CTA in first viewport
Social proofTestimonials, logos, case study links
Speed<3s load time or you lose 50% of visitors
SEOMeta tags, semantic HTML, structured data
ConversionOne clear CTA per page section

Output Template

ORCHESTRATOR REPORT
───────────────────
Project:        [Description]
Engine:         [Efficient / Robust β€” reversibility: ...]
Spec Status:    [Approved / Pending β€” design.md: ...]
Data Pipeline:  [DuckDB / Pandas / N/A β€” file size: ...]
Quality Gate:   [De-Sloppified: Y/N β€” coverage: X%]
Ship Threshold: [60% / 70% / 85% / 95% / 99%]

STATUS: [BUILDING / TESTING / SHIPPED]

Absorbed Protocols & Skills

Coding (6)

COD-107 (Spec-Driven Development), COD-108 (Semantic Search Standards), COD-110 (Structured Decoding), COD-112 (Stop Pattern), COD-415 (Spec-Driven Velocity), COD-900 (Project Scaffolding)

Absorbed Skills

  • data-analysis β†’ DuckDB-powered large file analytics
  • academic-delivery β†’ 8-step pipeline for academic deliverables
  • academic-humanizer β†’ AI detection bypass rewriting
  • spec-driven-dev β†’ Interrogation β†’ design.md β†’ build
  • statistical-analysis β†’ SPSS/R/Python statistical pipelines

Key Case Studies

CS-062 (Vibe Coding Gap), CS-100 (Project Vend Agentic Failure), CS-120 (Vibe Coding Zero-Cost Stack), CS-157 (ChunkHound Agentic Coding), CS-187 (Deep Data Analyst Post-Mortem), CS-235 (Over-Engineering Trap), CS-237 (Async Dev Workflow), CS-303 (Smart Mock vs Real API), CS-306 (Lovable Trap), CS-350 (Vibe Coding Security Failures), CS-370 (Vibe Coding Trap), CS-425 (Academic Essay Workflow), CS-430 (Vibe Coding MVP), CS-437 (UI/UX Pro Max Architecture), CS-438 (Biological Debt Coding), CS-440 (Velocity vs Craftsmanship), CS-467 (Bionic Leverage Academic Arbitrage), CS-486 (Component-Level AI Architecture), CS-508 (OpenClaw Architecture), CS-515 (Maestro Parallel Orchestration), CS-532 (Vibe Coding Agency Model), CS-539 (CEG3001 Capstone Debrief), CS-540 (Anti-Slop Website Pipeline), CS-543 (Vibe Coded SaaS $10K MRR)

Failure Modes & Mitigations

FailureMitigation
Building without specNEVER proceed without design.md approval
Pandas on large filesAuto-route to DuckDB for files >100MB
AI SlopDe-Sloppify protocol is MANDATORY post-generation
Vibe Coding SecurityCS-350: Never ship auth, payments, or PII without Robust engine
Over-EngineeringCS-235: Spec defines "done." Don't gold-plate.
Render JitterExtract primitive deps for useEffect. Never pass object refs.

Validated Patterns (Empirical)

  • [V] DuckDB > Pandas: For files >100MB, DuckDB is 10-50x faster and doesn't crash. | Reapply: Every large data analysis.
  • [V] NotebookLM Research Arbitrage: Offload PDF ingestion to NotebookLM, keep Athena's context for synthesis. | Reapply: Every academic assignment.
  • [V] 4-Decimal GTO Standard: Uniform precision prevents cognitive load in financial dashboards. | Reapply: Every statistical display.
  • [V] Primitive Dependency Extraction: [data.currentRatio] instead of [data] stops React re-render loops. | Reapply: Every dynamic status component.
  • [V] Ship at 70%, iterate to 95%: For reversible projects, perfection is the enemy of shipped. | Reapply: Every portfolio/MVP build.

References

  • META_PATTERNS.md β€” MP-2 (RETO engine), MP-11 (Iteration Economy)
  • bionic-decision-engine β€” For build/buy/wait decisions

Related skills