[Good First Issue]: add docs/competitors/sakana-fugu.md profile for the multi-agent orchestrator

[claude]

What

Add a new docs/competitors/sakana-fugu.md profile for Sakana
Fugu — Sakana AI's multi-agent orchestration model family released
2026-06-22 (Fugu + Fugu-Ultra), pitched as a "Conductor + Trinity"
agent stack. Then link the new profile from
docs/competitors/README.md. Mirror
the shape of docs/competitors/aider.md
or docs/competitors/cline.md. Pure docs.

Why

Multi-agent orchestration is shep's home territory — the project
README calls the product "an autonomous AI-native SDLC platform that
runs parallel AI agents in isolated worktrees." When a new
orchestration model lands from a high-credibility maker, our
competitors README should profile it so contributors can frame what's
already different and what's the same.

• Sakana Fugu — multi-agent orchestration model that matches Fable
  5 on quality (2026-06-22). Sakana AI's first dedicated
  multi-agent orchestrator, claimed to match Anthropic Fable 5
  quality on coding benchmarks. Trinity (planner) + Conductor
  (executor) architecture. ICLR 2026 paper.
• Qwen-AgentWorld — language world models that simulate seven
  agent domains (2026-06-23). Open-weights complement to Fugu in
  the same week.

The maintainer's existing project README directly invites comparison
against tools "like Cursor, Claude Code, Aider, Cline, OpenCode,
Codex, Windsurf, Continue, Codium" (see prompts/generate-good-first-issues.md line 22). Fugu is the new entrant in
that conversation this week.

Where

• docs/competitors/sakana-fugu.md — NEW FILE. Mirror docs/competitors/aider.md section-for-section: title + tagline blockquote, identity table (Website / Paper / Tagline / Type / Pricing / License), "What It Does" with bulleted Key Features (Trinity/Conductor split, multi-agent orchestration, ICLR 2026 paper, Japan-based maker), "How Shep Compares" table, "What We Respect", "Where Shep Differs", "Sources:" footer.
• docs/competitors/README.md:38-50 — refresh the "Agent-agnostic" row of the Capability matrix so that the listed "Closest Alternative" mentions Sakana Fugu's multi-agent split as a model-level analogue to shep's process-level orchestration (one sentence, no scope creep).
• docs/competitors/README.md:53-75 — add a new section heading Multi-Agent Orchestration Models with a single-row table linking to sakana-fugu.md. (Do not lump Fugu into "Open Source CLI Tools" — it's a model + SDK, not a CLI, and giving it its own section keeps future entries clean.)
• docs/competitors/README.md:90-103 — add one bullet to "What We Learned" crediting Sakana Fugu with formalizing the planner/executor split.
• Source material to quote verbatim: Sakana AI's Fugu release page and the ai-tldr summary item sakana-fugu (2026-06-22). Do not invent benchmarks.

Acceptance criteria

• New file docs/competitors/sakana-fugu.md follows the same section ordering as docs/competitors/aider.md. No invented new headings.
• The "How Shep Compares" row labels match those used in docs/competitors/cline.md (Interface / Focus / Requirements / Planning / Approval / Parallel features / CI integration / Dashboard).
• All factual claims (Trinity/Conductor naming, ICLR 2026 paper, release date 2026-06-22, "matches Fable 5 on quality" framing) cite a real source URL in the footer.
• docs/competitors/README.md adds: a new Multi-Agent Orchestration Models section linking to sakana-fugu.md; one refreshed cell in the Capability matrix's "Agent-agnostic" row; one new bullet in "What We Learned".
• No edits outside docs/competitors/. No code changes, no image additions.
• pnpm validate passes.

Hints for a first-time contributor

• Use docs/competitors/aider.md as a structural template — it's the closest existing profile for a category that isn't a CLI tool.
• The official Fugu release page is the primary source. If a fact isn't on that page or in the ICLR 2026 paper, leave it out.
• Don't claim shep ships Fugu or supports it as a provider — only that Fugu is a peer in the multi-agent orchestration space and worth understanding.

Trend hook

Sakana Fugu — multi-agent orchestration model that matches Fable
5 on quality (2026-06-22) is the highest-credibility multi-agent
orchestration model to ship this week, and shep's whole product
story is multi-agent orchestration. Our competitors README should
say where Fugu fits.
Source: ai-tldr 2026-06-22 — sakana-fugu.

---

Auto-generated by .github/workflows/generate-good-first-issues.yml
on 2026-06-26. Reply with @claude if anything is unclear — a
maintainer will review.