Skip to content

docs: competitive teardown — Symphony vs TaskYou#602

Open
bborn wants to merge 1 commit into
mainfrom
task/4360-compare-symphony
Open

docs: competitive teardown — Symphony vs TaskYou#602
bborn wants to merge 1 commit into
mainfrom
task/4360-compare-symphony

Conversation

@bborn

@bborn bborn commented Jun 13, 2026

Copy link
Copy Markdown
Owner

What

Comparison of Symphony (anantjain-xyz/symphony-rust) against TaskYou, per task #4360. Adds docs/comparisons/symphony.md.

The brief asked two things: what should we copy/borrow, and is their GUI built better than ours.

Key findings

On the GUI question — no, their stack is essentially identical to ours. Both are Tauri 2 + React + TypeScript (+ Tailwind/shadcn-style + SSE). We are not behind on GUI technology. The real difference is posture: Symphony's GUI is the only interface, while ours is one of three faces of the same engine. Our TUI-first + 100%-scriptable-CLI + MCP story is a genuine differentiator we should not dilute. Re-platforming the desktop GUI is explicitly not recommended.

What's worth borrowing (all engine-level features that benefit TUI/CLI/GUI at once):

  1. Automatic retry with exponential backoff + error context — our retry is manual-only (internal/ui/retry.go), which undercuts our unattended-execution story. Biggest gap.
  2. AI-provider rate-limit + token-usage awareness — we only track GitHub API limits, nothing about the agent provider's limits/tokens.
  3. Editable prompt templates with a placeholder reference panel — ours are hardcoded in Go.
  4. "Explicit user intent is never silently overridden" routing principle (not the Linear-specific feature).
  5. Keychain-backed secret storage on macOS as we add integration credentials.

Our moats (documented so we don't regress): TUI-first, scriptable CLI/MCP, local-first (no SaaS coupling), 6 executors vs 2, worktrees vs full clones, SSH + browser access, project-context caching, extensions ecosystem.

Files

  • docs/comparisons/symphony.md (new)

Follow-ups

The doc ends with a ranked, standalone-task list (auto-retry → rate-limit tracking → prompt templates → retry-queue view → keychain). These are recommendations, not implemented here — this PR is the analysis only.

🤖 Generated with Claude Code

Compares anantjain-xyz/symphony-rust against ty across product posture,
GUI architecture, and feature gaps. Key findings:

- GUI stacks are near-identical (both Tauri 2 + React + TS); no rearchitecture
  needed. Our TUI-first / fully-scriptable posture is the real differentiator.
- Highest-value borrowables are unattended-run robustness features:
  automatic exponential-backoff retry with error context, AI-provider
  rate-limit/token awareness, and editable prompt templates.
- Documents our moats (TUI-first, scriptable CLI/MCP, worktrees, 6 executors,
  SSH, project-context caching, extensions) to avoid regressing them.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant