improv.sh vs Unspaghettit

improv.sh and Unspaghettit are both cli coding agents tracked by AIDiveForge. Below is a side-by-side comparison of pricing, capabilities, platforms, and ownership — sourced from each tool's live website and verified before publishing.

improv.sh

improv operates as a task harness: the @im call pulls repo context, detects your test commands, writes acceptance criteria, and packages shell validation steps into one spec the agent can implement on turn one. The loop infrastructure is the distinguishing piece — judges run your actual exit-code commands (npm test, typecheck, build), so done means your tests pass, not that the agent says it's done. The tool installs locally via curl with no external API keys required, and the Chrome extension brings the same engine into web-based chat interfaces. The 920-skill library and daily auto-research loop suggest the routing layer will keep growing — but the page offers no independent benchmarks to validate the token-savings figures cited.

Unspaghettit

Orbit wraps each coding-agent invocation in a bounded loop: it selects a dependency-ordered task from a backlog, runs the agent, then gates advancement on passing tests, lint, and type checks — not on the agent's self-report. Every run writes structured JSON artifacts and a human-readable progress log, so you can inspect what changed and why a task closed or stalled. The deterministic replay demo runs without an API key, which means you can verify the harness behavior before committing any agent credits. The ceiling appears when your workflow needs anything beyond CLI-compatible agents — there is no API and no visual interface.

Attribute	improv.sh	Unspaghettit
Pricing	Free	Free
Free trial	No	No
Open source	Yes	Yes
Has API	No	No
Self-hosted option	Yes	Yes
Platforms	VS Code, Cursor, Claude Code, terminal, Chrome	Linux, macOS, Windows (Python-based)
Pros	Repo-aware spec compilation pulls your actual test commands and package scripts into the task before the agent starts, which means the agent implements against your real constraints instead of inventing them mid-run. Exit-code judges close the loop on real shell commands — npm test, typecheck, build — so you are not relying on the agent's self-assessment of whether it finished. Task memory persisted under .improv/tasks/ survives session boundaries, so an agent restarted mid-task picks up status and spec instead of starting the discovery cycle again. Local-first install with no external API keys required, which means the harness runs in air-gapped or locked-down environments where cloud tooling is blocked. Chrome extension and VS Code/Cursor Marketplace extension share the same local engine, so the spec compilation and judge loop work whether you are in the IDE or a browser-based chat interface — without switching tabs.	Proof-gated task closure — tests, lint, and type checks must pass before an orbit advances — which means you stop shipping agent output that looked correct in the diff but broke downstream. Structured JSON artifacts on every run (agent-result.json, evaluation.json, review.json, progress.md), so debugging a failed orbit means reading a file rather than reconstructing what the agent did from memory. Agent-neutral adapter contract, so you can run Claude and Codex against the same task backlog and compare evaluation scores instead of arguing from anecdotes. Deterministic replay demo requires no API key, which means the harness itself is verifiable in CI before any live agent is connected — reducing the risk of paying for agent credits on a broken setup. Dependency-aware backlog selection keeps each agent invocation scoped to one task, which means you avoid the compounding errors that come from letting an agent chain across unverified intermediate states.
Cons	Task state is written to .improv/tasks/ on the local machine. Teams with more than one developer working the same codebase have no shared task state — there is no sync layer described on the page — so parallel agent runs on different machines produce divergent task records with no reconciliation path. The tool exposes no API surface, so teams that want to trigger improv from a CI pipeline or wrap it in a custom orchestration layer cannot. Teams hitting this wall move to harness frameworks that expose programmatic interfaces — at which point they are maintaining the prompt compilation logic themselves. The token-savings figures on the page (~613 tokens median) are vendor-reported with no independent reproduction methodology described. Teams making adoption decisions based on cost reduction should treat these numbers as illustrative until they run their own baseline comparison. Chrome extension installation requires either the Chrome Web Store or a manual sideload script — neither path is available in Firefox or Safari. Teams on non-Chromium browsers are limited to the terminal install, losing the browser chat integration entirely.	Orbit requires agents that speak JSON over CLI. Agents with proprietary APIs, browser-based interfaces, or non-CLI outputs cannot be connected without writing a custom adapter — a task the docs acknowledge but leave entirely to the contributor. There is no hosted option, no REST API, and no web interface. Teams that need to hand off agent monitoring to non-engineering stakeholders, integrate Orbit into an existing SaaS workflow, or run it without local infrastructure have no path forward within the current scope. The harness assumes a test suite exists and is the source of truth for correctness. Repositories without meaningful test coverage get validation gates that pass trivially, which defeats the proof model entirely — at that point teams are back to trusting agent self-reports. Teams that need agents running in parallel across multiple tasks, conditional branching based on intermediate outputs, or cross-agent handoffs will hit the single-orbit-at-a-time design ceiling quickly. When that happens, the documented response is to build on top of Orbit or move to a more full-featured orchestration layer — at which point Orbit becomes a sub-component rather than the primary harness.

Bottom line

improv.sh and Unspaghettit are closely matched on pricing model, openness, and API availability — pick by feature set and platform support in the table above.

Frequently asked questions

What is the difference between improv.sh and Unspaghettit?

improv.sh is Free and open source, while Unspaghettit is Free and open source. Compare pricing, free trial, API, platforms, and pros/cons in the table above on AIDiveForge.

Is improv.sh better than Unspaghettit?

It depends on your workflow. Use the side-by-side attributes (pricing, open source, API, self-hosted, platforms) to decide. AIDiveForge does not rank a universal winner — we publish verified facts so you can choose.