research-loop

Use when user mentions autonomous iteration, metric-driven optimization, $research-loop plan, $research-loop debug, $research-loop fix, $research-loop security, $research-loop ship, $research-loop scenario, $research-loop predict, $research-loop learn, $research-loop reason, $research-loop probe, or mentions "research-loop" with a goal/metric. Autonomous Goal-directed Iteration — apply Karpathy's autoresearch principles: modify, verify, keep/discard, repeat. Supports bounded mode via Iterations: N inline config.

Mandatory activation layer — loads on any conversation start. Establishes skill-loading protocol, Red Flags, priority rules, and HARD-GATE enforcement for all research-loop skills.

Run a thorough, multi-phase deep research investigation on a topic with subagent dispatch, provenance tracking, and integrity verification.

Experiment sandbox execution for Research Loop. Supports four modes: local (venv), Docker (isolated containers), SSH remote (GPU compute on servers), and Colab (Google Drive bridge). Provides experiment harness templates, code validation, metric collection, deterministic seeding, and compute budget enforcement. Use before running experiments generated by the paper-pipeline.

Use when user wants to explore a topic, find papers, map a field, or understand the research landscape. Not for explaining concepts — use learn for that.

Publication-quality figure generation for research papers. Decision agent selects figure type (code plot vs architecture diagram). Generates Matplotlib/Seaborn code for quantitative figures with iterative improvement loop. Style-matches conference templates (NeurIPS, ICML, ICLR). Use when the paper-pipeline reaches the figure generation phase, or when a user requests figures for an existing draft.

Use when user asks what something means, says "explain", "I don't understand", "teach me", "what is X", or asks about a term mid-session.

Run a structured literature review on a topic using parallel search, evidence tables with quality scoring, and primary-source synthesis.

End-to-end paper generation pipeline ported from AutoResearchClaw (Aiming Lab). 14 phases covering topic initiation through export/publish, with human- in-the-loop gates and quality gating at each handoff. Use this when the user wants a full paper pipeline run — topic to submission-ready manuscript. Delegates to researcher/reviewer/writer/verifier subagents for stage execution and to autonomous-iteration for experiment optimization loops.

Plan and execute a structured replication workflow for a paper, claim, or benchmark with environment selection and integrity checks.

Use when user has a hypothesis and a repo and wants to run experiments. Triggered by "start experiments", "run the loop", "test this hypothesis".

Use when user asks "where are we", "show me the status", "what do we have", "what's pending", "what should I do next", or wants to see the research decision tree.

Use when user wants to test multiple angles on an idea, run parallel hypothesis lanes, or explore a gap from different entry points.

Use when experiments are running or just completed, or user shares results and wants to decide what to do next.

Use when user wants to explore a topic, find papers, map a field, or understand the research landscape. Not for explaining concepts — use learn for that.

Use when user wants to find gaps, evaluate ideas, or decide what's worth pursuing. Triggered by "what hasn't been tried", "is this a good idea", "find the gap".

Use when user asks what something means, says "explain", "I don't understand", "teach me", "what is X", or asks about a term mid-session.

Use when user has a hypothesis and a repo and wants to run experiments. Triggered by "start experiments", "run the loop", "test this hypothesis".

Use when user has a selected hypothesis or route and needs a concrete execution plan broken into tasks. Triggered by "what are the next steps", "how do I start", "plan this out".

Use when user mentions research, a topic, papers, experiments, gaps, or hypotheses. Entry point — load this first.