Slash Commands

Complete reference for all REPL slash commands available in Feynman.

Slash commands are available inside the Feynman REPL. They map to research workflows, project management tools, agent delegation utilities, and bundled package capabilities. Type /help inside the REPL for the live command list, which reflects your installed packages and may include additional commands.

Research workflows

These are the primary commands you will use day-to-day. Each workflow dispatches one or more specialized subagents (researcher, reviewer, writer, verifier) depending on the task.

Command	Args	Description	Produces
/deepresearch <topic>	<topic>	Run a thorough, source-heavy investigation on a topic and produce a durable research brief with inline citations	outputs/<slug>.md + outputs/<slug>.provenance.md
/lit <topic>	<topic>	Run a literature review using paper search and primary-source synthesis	outputs/<slug>.md + provenance sidecar
/review <artifact>	<artifact>	Simulate an AI research peer review with likely objections, severity grading, and a concrete revision plan	outputs/<slug>-review.md
/audit <item>	<item>	Compare a paper’s claims against its public codebase and identify mismatches, omissions, and reproducibility risks	outputs/<slug>-audit.md
/replicate <paper>	<paper>	Plan or execute a replication workflow for a paper, claim, or benchmark	Replication plan, scripts, and results on disk
/compare <topic>	<topic>	Compare multiple sources on a topic and produce a source-grounded matrix of agreements, disagreements, and confidence	outputs/<slug>-comparison.md
/draft <topic>	<topic>	Turn research findings into a polished paper-style draft with equations, sections, and explicit claims	papers/<slug>.md
/autoresearch <idea>	<idea>	Autonomous experiment loop — try ideas, measure results, keep what works, discard what doesn’t, repeat	Iteration log and autoresearch.md session state
/watch <topic>	<topic>	Set up a recurring or deferred research watch on a topic, company, paper area, or product surface	outputs/<slug>-baseline.md + scheduled follow-up

All research workflow slash commands can also be invoked directly from the CLI — for example, feynman deepresearch "topic". See CLI Commands for details.

Workflow details

/deepresearch — Deep research brief

Runs a multi-round research sweep using parallel researcher subagents, then synthesizes findings into a cited and verified brief. Each run produces a plan artifact, intermediate research files, a draft, and a final cited output with a provenance sidecar.

The workflow always presents a plan and asks for confirmation before starting. For unattended execution, ask Feynman to run it asynchronously.

/deepresearch mechanistic interpretability in transformers

/lit — Literature review

Searches papers and primary sources, then produces a structured literature review separating consensus, disagreements, and open questions. Uses the researcher, verifier, and reviewer subagents.

/lit diffusion models for protein structure prediction

/review — Peer review simulation

Evaluates an artifact (paper, draft, or output file) against research quality criteria: novelty, empirical rigor, baselines, and reproducibility. Returns severity-graded feedback and a revision plan.

/review outputs/my-draft.md

/audit — Claim vs. code audit

Compares a paper’s stated methods, defaults, and reported metrics against its actual public codebase. Flags missing code, mismatches, ambiguous defaults, and reproduction risks.

/audit 2401.12345
/audit https://arxiv.org/abs/2401.12345

/replicate — Replication workflow

Extracts implementation details from a paper and produces a structured replication plan. If you choose an execution environment (local, venv, Docker, Modal, RunPod), it implements and runs the steps there. Requires explicit environment confirmation before executing any code.

/replicate GPT-4 MMLU few-shot benchmark

/compare — Source comparison matrix

Gathers source material across multiple papers, posts, or repositories and builds a comparison matrix covering key claims, evidence types, caveats, and confidence levels. Generates charts for quantitative comparisons.

/compare LoRA vs full fine-tuning for instruction following

/draft — Paper-style draft

Writes a structured paper from research findings. Includes title, abstract, problem statement, related work, method or synthesis, evidence, limitations, and conclusion. Uses LaTeX for equations and pi-charts for data figures.

/draft scaling laws for retrieval-augmented generation

/autoresearch — Autonomous experiment loop

Iteratively tries ideas against a benchmark metric, keeps improvements, reverts regressions, and logs every iteration. Collects metric name, benchmark command, scope, and max iterations before starting. Supports local, venv, Docker, Modal, and RunPod environments.

/autoresearch minimize test suite runtime
/autoresearch off       # Stop the loop, keep data
/autoresearch clear     # Delete all state

/watch — Recurring research monitor

Runs a baseline sweep on a topic and schedules a recurring or deferred follow-up using schedule_prompt. Useful for tracking a research area, company, or paper cluster over time.

/watch llm reasoning benchmark progress

Project and session

Command	Description
/help	Show grouped Feynman commands and prefill the editor with a selected command
/init	Bootstrap AGENTS.md and session-log folders for a new research project
/outputs	Browse all research artifacts (papers, outputs, experiments, notes)
/jobs	Inspect active background research work — running processes and scheduled follow-ups
/log	Write a durable session log with completed work, findings, open questions, and next steps

Agents and delegation

These commands are provided by the Pi agent runtime and enable multi-agent workflows inside the REPL.

Command	Usage	Description
/agents	/agents	List available subagents
/run	/run <agent> <task>	Delegate a task to a named subagent
/chain	/chain agent1 -> agent2	Run agents in sequence, passing output forward
/parallel	/parallel agent1 -> agent2	Run agents concurrently

Feynman ships four bundled research subagents: researcher, reviewer, writer, and verifier. These are defined in .feynman/agents/ and synced at startup.

/run researcher "Gather recent papers on speculative decoding"
/chain researcher -> writer
/parallel researcher -> reviewer

Bundled package commands

These commands are provided by core Pi packages bundled with every Feynman installation.

Command	Usage	Description
/ps	/ps	List active background processes and experiments
/schedule-prompt	/schedule-prompt	Schedule a deferred or recurring prompt
/search	/search	Search prior session transcripts (requires session-search package)
/preview	/preview	Preview the current artifact as rendered HTML or PDF
/new	/new	Start a new session
/quit	/quit	Exit the REPL
/exit	/exit	Exit the REPL

/preview requires pandoc. Run feynman --setup-preview or feynman setup to install it.