Skip to content
UsecaseBoard
Saved0
← All use cases
hnmulti-agent

Sakana Agentic Scientist v3: coding agents as experiment loop

At a glance

Open pipeline turns a task.md into idea→experiments→ablations→sealed eval→paper, orchestrating coding agents (Docker/yolo by default) for automated scientific co-discovery. This workflow is listed under multi-agent, science, experimentation, coding-agents, orchestration.

Agents / tools
Sakana Agentic Scientist, agsci, coding-agent CLIs, Docker

Building a tool for these workflows? Sponsor / inquire.

Save ideas in this browser. Upvotes count across the board.

Prompt excerpt

Turn a task description into experiments, ablations, and a paper, driven by coding agents.

Related workflows

Explore all →
productcoding-agents

Autonomous Harness: desk command center for every coding agent and machine

Autonomous Labs’ Harness (Product Hunt #2 on Oct 16; open-sourced desktop/mobile/firmware) is a round-display desk device plus native app that shows Claude Code, Codex, Cursor, OpenCode, Pi, Hermes, Copilot and more across laptop/server/cloud—voice a task, see runtime and spend at a glance, get a quiet ping when work returns. E2E-encrypted relay; agents keep running if the UI closes. Domain harnesses extend beyond code (Blender, CircuitJS, MuJoCo, Godot).

HarnessClaude CodeCodex+4
Source ↗
productmulti-agent

Bandito: Mac crew that runs Claude Code/Codex/Grok on your servers

Native macOS app orchestrates official Claude Code, Codex and Grok CLIs as a role-based crew (lead/builder/reviewer/docs/auditor/on-call) on your Mac or VPS with schedules, one-click risky-command approvals (Telegram too), and server screen/browser/terminals—no Bandito cloud for your code.

BanditoClaude CodeCodex+1
Source ↗
blogMeta-Muse

Muse Code: fan out parallel subagents into isolated git worktrees

Meta’s Muse Code co-trained agent (with Muse Spark) fans work out automatically: parent spawns write-capable children, each in its own git worktree so parallel edits never collide; background observer agents stay live for memory/skill/goal/verification. Fully auditable event log; workers + reviewers by default. Install: curl -fsSL https://dev.meta.ai/install.sh | bash.

Muse CodeMeta MuseMuse Spark+1
Source ↗
productcoding-agent

Tern: free native fleet IDE—one worktree per agent, one review queue

Tern is a free closed-source Mac-native (CLI/TUI also on Linux) control plane for 16 coding-agent CLIs: each task gets its own git worktree/branch, changes land in one review queue with comments routed back to the agent, conflict radar, multi-step runs with gates, phone approvals without Tern servers, and a daemon that keeps agents alive after you close the window.

TernClaude CodeCodex+3
Source ↗