SYSTEM RUNTIME: POSIX • 127 ACTIVE DIRECTIVES • DETERMINISTIC PROTOCOL

Autonomous Skill Tree & Directives for Claude Code and Antigravity IDE

Q: How are these directives installed on local machines?

The install script places the skills directory tree into ~/.gemini/config/skills/ and appends the 3-step operational protocol to ~/.gemini/config/GEMINI.md.

127 modular system directives. Deterministic execution protocols. Zero LLM hallucinations. Enforces 10-year senior engineering and local search architecture on every session.

bash — deployment execution POSIX bash/zsh

$git clone https://github.com/jahidulislamseo/antigravity-ai-skills.git && cd antigravity-ai-skills && ./install.sh

127 Skills Packaged

39 Local SEO Engines

66 Engineering Stacks

MIT Open Source

Execution Diagnostics

Standard LLM Output vs. Antigravity Directives

Comparison of failure modes in unconstrained generative AI models versus deterministic directive execution.

Audit Vector	Unconstrained Default LLM	Antigravity AI Directives
Local SEO Auditing	Superficial "Post photos weekly and collect positive reviews on your Google Business Profile."	Operational Audits primary category dilution, missing secondary services, geoCoordinates, and review velocity formulas.
Empirical Claims	Hallucinated Claims ungrounded metrics (e.g. "This optimization will boost conversions by 35%").	Grounded Explicitly bans fabricated percentage claims. Missing data returns "Data Unavailable - Required: [Metric]".
Code Generation	Incomplete Outputs fragile boilerplate littered with `// TODO: implement logic` comments.	Complete Generates production-first code with runtime type safety, error boundaries, and unit test coverage.
Deliverable Gating	Unsolicited Buries the user in unrequested 15-page slide outlines when asking for a simple technical diagnosis.	Gated Strict anti-fluff protocol. Delivers direct code and concrete fixes without unrequested report decks.

Protocol Architecture

The Deterministic 3-Step Sequence

Every prompt passed to the agent passes through this immutable execution pipeline.

STEP 01

ROCS-EO Prompt Calibration

The agent first converts raw instructions into a 10-year senior prompt schema defining Role, Objective, Constraints, System process, Evidence schema, and Output contract.

STEP 02

Autonomous Skill Chaining

Chains the required modular skill folders (e.g., gbp-optimization ➔ local-schema ➔ review-management) from ~/.gemini/config/skills/.

STEP 03

Strict Deliverable Gating

Outputs concrete fixes and validated code immediately. Formal executive audits and visual slide decks are produced strictly when explicitly ordered by the user.

Directive Index

Directory of 127 Specialized Directives

Query and filter the active skill tree across local search, organic content, systems engineering, and client deliverables.

Showing 127 active directives:

Empirical Case

Case Evidence: Amethyst Dental (Victorville, CA)

Demonstration of physical category dilution isolation versus default generic AI advice.

P0 Algorithmic Penalty Diagnosis

Default AI advised "updating social links". The Antigravity local diagnostic protocol identified the operational root cause:

—
Category Dilution: 8 divergent specialties attached to a general dentistry practice.
—
Intent Mismatch: Missing Emergency dental service secondary category despite advertising emergency slots.
—
Engineered Solution: Pruned categories to 1 primary + 3 secondaries, and delivered validated Dentist JSON-LD schema.

// Production Validated Output: { "@context": "https://schema.org", "@type": "Dentist", "@id": "https://amethystdental.com/#dentist", "name": "Amethyst Dental Care", "telephone": "+1-760-241-1111", "address": { "@type": "PostalAddress", "streetAddress": "12345 Amethyst Rd", "addressLocality": "Victorville", "addressRegion": "CA", "postalCode": "92392" }, "geo": { "@type": "GeoCoordinates", "latitude": 34.5362, "longitude": -117.2928 }, "medicalSpecialty": "Dentistry" }

Documentation

Frequently Asked Technical Questions

Technical architecture, runtime execution, and integration pathways.

The install script places the skills directory tree into ~/.gemini/config/skills/ and appends the 3-step operational protocol to ~/.gemini/config/GEMINI.md. Host agents load them whenever task patterns match.

The operational directive explicitly bans LLM marker vocabulary, negative parallelism, symmetrical adjective stacking, and unsupported causal metrics. Agents must provide operational mechanisms instead of adjectives.

Anthropic Claude Code, Google Antigravity IDE, Cursor via custom rules, and any MCP client capable of reading structured SKILL.md documents with YAML frontmatter.