autonomous-llm-research

Automating literature review for LLM post-training research involves systematizing literature search and gap identification. This Skill handles that autonomously, converting raw research ideas into structured plans and tracking evaluation metrics.

Planning and checkpointing LLM experiments for resumable research requires building reusable plans and persisting state. This Skill creates durable research loops with persistent state stored under a .hermes-research/ directory.

Generating and evaluating hypotheses for autonomous LLM research is handled by suggesting hypotheses and tracking evaluation metrics. This allows you to turn vague ideas into formal reports through a structured, automated evaluation process.

Yes, you can use this Skill to turn a vague research idea into a formal report. You provide a high-level research objective, and it initializes a durable post-training research loop that outputs structured plans, experiments, and final reports.

Autonomous LLM research with this Skill requires no external dependencies or components. It operates independently to manage stateful project creation and integrates directly with the management layer via research_manager actions.