Core Agent Skills
Relevant source files
The following files were used as context for generating this wiki page:
Core Agent Skills represent the specialized operational capabilities and automated workflows utilized by autonomous agents within the ÆtherLight platform. These skills enable agents to perform complex engineering tasks—ranging from project initialization and codebase analysis to infrastructure orchestration and automated documentation—while maintaining strict adherence to project standards and hierarchical context loading patterns.
These skills are integrated into the Agent Context Architecture, ensuring that each agent type (e.g., Infrastructure, Documentation, Review) has access to the specific tools and procedural knowledge required for their role without overwhelming the token budget.
Sources: internal/agents/README.md, internal/agents/documentation-agent-context.md:16-30, internal/agents/infrastructure-agent-context.md:10-23
Initialization and Project Setup
The initialization skill is responsible for establishing the foundation of a new project or integrating the engineering platform into an existing one. It automates the creation of standardized directory structures, configuration files, and git workflows.
Process Flow
The initialization process follows a structured sequence to ensure environmental consistency:
The initialization skill detects project types (Node.js, Rust, Python, etc.) and establishes core directories such as .claude/skills, sprints/, and .aetherlight/patterns/.
Sources: skills/initialize/SKILL.md:21-65, skills/initialize/SKILL.md:68-80
Key Components
| Component | Description |
|---|---|
| CLAUDE.md | The primary AI context file containing project instructions and standards. |
| settings.json | Defines global workflow rules, such as mandatory PRs and semantic commits. |
| Pattern Library | A repository of reusable templates for Auth, Database, and API logic. |
| SPRINT_*.toml | Configuration files for managing task lifecycle and agent assignments. |
Sources: skills/initialize/SKILL.md:143-162, skills/initialize/SKILL.md:329-354
Code Analysis and Quality Enforcement
The code-analyze skill enables agents to perform standardized reviews of the codebase to identify technical debt, security vulnerabilities, and pattern deviations.
Analysis Phases
- Structure Analysis: Mapping entry points and codebase metrics using tools like
cloc. - Pattern Detection: Identifying deprecated logic or high-complexity functions.
- Security/Performance Scanning: Checking for exposed secrets or synchronous bottlenecks in critical paths.
Sources: skills/code-analyze/SKILL.md:27-55, internal/agents/review-agent-context.md:80-105
Automated Reporting
The skill generates structured reports that categorize findings into "Must Fix" (Blocking), "Should Fix" (Warning), and "Consider Fixing" (Info).
Infrastructure and Service Orchestration
The Infrastructure Agent utilizes specialized skills to design and implement middleware services, primarily focusing on VS Code extension components and service orchestration.
Service Architecture
Infrastructure skills prioritize high-performance service response times (e.g., workflow checks <500ms) and strict adherence to the MiddlewareLogger pattern for structured observability.
Sources: internal/agents/infrastructure-agent-context.md:10-23, internal/agents/infrastructure-agent-context.md:95-102
Pre-Task Analysis Protocol
Before implementing any infrastructure change, agents must execute a mandatory analysis checklist:
| Step | Requirement |
|---|---|
| Agent Verification | Confirm agent assignment matches task domain. |
| Tech Stack Check | Verify VS Code API and Node.js version compatibility. |
| Integration Analysis | Identify existing services (e.g., ConfidenceScorer, TestValidator). |
| Library Selection | Prefer Node.js built-ins over runtime npm dependencies. |
Sources: internal/agents/infrastructure-agent-context.md:33-46, internal/agents/infrastructure-agent-context.md:155-170
Documentation and Chain of Thought
The Documentation Agent ensures that all project components adhere to the Chain of Thought (CoT) documentation standard. This skill is critical for knowledge preservation and AI interpretability.
Required Documentation Elements
Every function, class, or module must include:
- DESIGN DECISION: The approach taken.
- WHY: The reasoning behind the decision.
- REASONING CHAIN: A step-by-step numbered logic list.
- PATTERN: Reference to the applicable platform pattern.
Sources: internal/agents/documentation-agent-context.md:24-35
Validation Workflow
Lifecycle and Task Management
All agents utilize a shared Sprint Task Lifecycle Protocol to synchronize state across the platform. This ensures that the Sprint Panel and other UI components reflect real-time progress.
Lifecycle Commands
Agents execute specialized commands to update their status in the ACTIVE_SPRINT.toml:
/sprint-task-lifecycle start {TASK-ID}: Moves status toin_progress./sprint-task-lifecycle complete {TASK-ID}: Marks ascompletedwith a timestamp./sprint-task-lifecycle defer {TASK-ID}: Records deferral reasons and blockers.
Sources: internal/agents/documentation-agent-context.md:235-275, internal/agents/infrastructure-agent-context.md:49-88
Conclusion
Core Agent Skills provide the operational framework that allows the ÆtherLight platform to function autonomously. By standardizing tasks such as initialization, analysis, and documentation into repeatable, validated skills, the platform ensures that agent actions are predictable, performant, and compliant with established engineering patterns. This skill-based approach is the foundation for scaling multi-agent workflows in complex software environments.
Sources: internal/agents/README.md, internal/agents/documentation-agent-context.md:295-300