run-local

Orchestrate local AutoReview pipeline stages via Claude Code subagents with snapshot resume.

Updated Feb 24, 2026
One-click install
npx skills add https://github.com/mcleanT/AutoReview --skill run-local-mcleant
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: run-local
Source: https://github.com/mcleanT/AutoReview/tree/main/.claude/skills/run-local
Command: npx skills add https://github.com/mcleanT/AutoReview --skill run-local-mcleant

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Running the AutoReview pipeline locally without external LLM API keys is complex and error-prone; this Skill provides a single, auditable execution contract that orchestrates each pipeline stage via Claude Code subagents, enforces validation gates, and preserves resumable snapshots to avoid silent stage-skips and lost work.

Core Features & Use Cases

  • Mandatory configuration interview that collects topic, domain, depth, model/provider, output targets, and resume options before any execution.
  • Deterministic stage orchestration across 17 named stages (discovery, analysis, planning, enrichment, writing, assembly, polish) with per-stage models, tools, and validation gates.
  • Snapshotting and resume: save per-stage JSON snapshots, token and timing logs, and support resume-from-stage workflows for crash recovery and debugging.
  • Search and citation rules: mandates direct Python SearchAggregator use for pipeline searches, tiered citation planning, and strict citation validation during section writing.
  • Use case: run a fully autonomous literature review locally when API keys are unavailable, audit each stage, and resume work after interruptions.

Quick Start

Ask the skill to "Run the AutoReview pipeline locally on the topic the role of gut microbiome in neurodegenerative diseases and save outputs to output/."

Frequently Asked Questions about run-local

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run an automated literature review locally without external LLM API keys?

You can run an automated literature review locally without external API keys by orchestrating Claude Code subagents across each pipeline stage, enforcing per-stage validation gates, and saving intermediate outputs to snapshot files for resumable execution.

Can I resume an interrupted literature review pipeline from a specific stage?

Yes, you can resume an interrupted literature review pipeline from a specific stage using snapshot-based resume workflows that load previously saved per-stage JSON snapshots and token timing logs to continue execution without silent stage-skips or lost work.

What is the best way to debug a literature review pipeline stage by stage?

The best way to debug a literature review pipeline stage by stage is to use deterministic stage orchestration with per-stage validation gates and snapshot files, allowing you to isolate, inspect, and resume execution from individual stages like discovery, analysis, enrichment, or writing.

Do I need to configure specific parameters before starting a local literature review run?

Yes, a mandatory configuration interview is required before execution, collecting the research topic, domain, depth, model or provider, output targets, and resume options to ensure deterministic stage orchestration and proper search and citation validation.

How does citation validation work during automated literature review writing?

Citation validation during automated literature review writing works by mandating direct Python SearchAggregator usage for pipeline searches, applying tiered citation planning, and enforcing strict citation validation gates during the section writing and assembly stages.