SPACEWALK
Official@teamspwk · Seoul, Korea
AI Architecting, Landbook
Agent Skills by SPACEWALK
Showing 15 vetted skills indexed across 2 GitHub repositories.
TeamSPWKevaluator
Verify Nova code implementations through adversarial static and execution-based evaluation.
TeamSPWKjury
Orchestrate 3-way jury evaluations for code review, plan review, and security diagnosis.
TeamSPWKdeepplan
Generate CPS-aligned plans for complex tasks using Explorer triplets.
TeamSPWKwriting-nova-skill
Author Nova Skills with standardized frontmatter, sections, and trigger tests.
TeamSPWKfield-test
Coordinate field tests across repositories using isolated worktrees and structured feedback reports.
TeamSPWKworktree-setup
Create idempotent symlinks for env and secret files in new git worktrees.
TeamSPWKevolution
Orchestrate scanners, filters, proposals, and gate chains for Nova self-evolution.
TeamSPWKcontext-chain
Propagate session context across Nova sessions via NOVA-STATE.md state and event updates.
TeamSPWKux-audit
Run five parallel adversarial evaluators to identify UI/UX issues and update NOVA-STATE.md.
TeamSPWKorchestrator
Coordinate multi-agent development workflows across design, planning, implementation, and validation phases.
TeamSPWKstrategic-compact
Guide session context compression decisions after milestones or 70% token usage.
TeamSPWKclaude-filter
Audit prompts, CLAUDE.md, and agent SDK settings with five-axis PASS/WARN/FAIL scoring.
TeamSPWKpikes-filter
Evaluate source code against Rob Pike's five principles and output a structured review report.
TeamSPWKllm-review
Run external LLM reviews across Gemini and OpenAI with consistent JSON outputs.
TeamSPWKdeep-dive-task
Coordinate multi-AI consultations to produce implementation-ready TASK documents.
Frequently Asked Questions About SPACEWALK
FAQPage SchemaWhat specific tasks does SPACEWALK enable for engineering teams?▼
SPACEWALK enables adversarial static evaluation, multi-way jury code reviews, and structured task planning. It facilitates complex system evolution through gate chains, context compression, and multi-model review outputs, ensuring code implementations align with specific architectural principles and security requirements.
Which personas benefit most from these evaluation frameworks?▼
Senior software architects, security engineers, and technical leads benefit from these frameworks. The system is designed for professionals managing complex, multi-agent development environments who require rigorous, standardized validation of code, UI/UX audits, and strategic context management across large-scale repositories.
What are the prerequisites for implementing these evaluation modules?▼
Implementation requires a Git-based repository structure and the ability to manage isolated worktrees. Users must maintain standardized documentation formats, specifically NOVA-STATE.md and CLAUDE.md, to allow the evaluation modules to propagate context, perform state updates, and execute adversarial audits effectively.