Free · BYOK · Desktop + VS Code · Windows & Linux

See your app before it's built.
Critique the picture, not the prompt.

You can't describe an app you haven't seen — nobody can. Peregrine Code draws your idea first: screens, entities, flows, side by side. You fix it in plain words until it looks right — and the approved picture becomes a living spec that keeps the agent from wrecking your app at 70%. Built by the fastest coding agent we know how to make.

Download — it's free How visual-first works

Your API keys never leave your machine · No account, no credit packs · 2 parallel tasks free

Ideation — post-generator
Structure
Product — name · photo · price
Post — text · image · status
Screens: Home · Editor
Screens: + Scheduler — why: post when the audience is online
Flow: Home → Editor → publish
Mockup
CONSTITUTION.md SPEC.md — EARS + your “why” STRUCTURE.md DESIGN.md tasks/phase-1.md — verify: npm test
I want a post generator with product photos
Drew it: 2 entities, 2 screens, 1 flow. Critique in words.
Nav into a left sidebar. Add a scheduler — posts should go out when the audience is online.
Updated both panes. Approved → the picture just became your living spec.

Real flow from the editor: idea → structure + mockup → critique in words → spec files born from the approved picture.

The picture becomes the spec.

Requirements engineering has a name for how humans actually work: “I'll know it when I see it.” It's easier to critique something in front of you than to describe something brand new — so Peregrine puts the picture first and makes the spec a byproduct of your approval.

01 Drop the idea

One sentence in chat. The agent drafts entities, screens and flows — a structure you can point at, not a wall of questions.

02 Critique in words

“Sidebar left. Add a scheduler. That relation is wrong.” Structure and mockup are projections of one model — a fix anywhere updates both. Undo included.

03 Approve → spec is born

CONSTITUTION, EARS requirements carrying your “why”, structure, design, phased tasks with verify lines. Separate files, in your repo, yours to edit.

04 It stays alive

The agent reads the constitution every single turn. Code moved ahead? One click re-checks the spec against reality and shows you the drift.

Why apps don't stall at 70%

No doom loop. No sea of markdown.

Vibe tools build instantly — then every new change breaks two old ones, because nothing remembers what you agreed on. Spec tools make you author documents before you've seen anything. Peregrine takes the third path.

The 70% wall, solved upstream

“You weren't clear about what you wanted” — because you couldn't be, yet. Visual agreement before code means the last 30% was designed in, not prompted around.

A constitution the agent can't ignore

Invariants, decisions and your “why” are injected into every agent turn. If a request conflicts with an invariant, the agent stops and says so — no silent violations, no forgotten decisions.

Spec benefits without spec ceremony

No “sea of markdown” you author by hand and watch drift. The spec is generated from what you approved, tasks come phased with executable verify lines, and a drift check re-syncs it from code in one click.

Already have a project?

Reverse Harness reads any existing codebase — yours, inherited, or a vibe-coded export that hit the wall — and reconstructs the same living spec from it. Every claim marked [certain] or [assumption]; when code and docs disagree, code wins.

Parallel where your code allows it.
Sequential only where it must be.

Other agents execute a plan step by step. Peregrine asks its own MCP dependency graph which steps are truly independent — and runs them as concurrent implementer agents, each locked to its own files.

fix-rate-limiting · 3 scouts · 9-step plan · 3 independent groups done in 84s — sequential estimate 4m 20s
scouts ×3
brief
planner
plan + graph
implementer 1
rateLimit.ts
review
implementer 2
auth actions
review
implementer 3
url guard
review
integration
wire + build
final review
quality gate

Cheap scouts hand the frontier planner a compact brief instead of raw files. Steps touching related modules are grouped via the import graph (get_file_relations), and each group is reviewed the moment it finishes — while its siblings are still implementing.

implementers in parallel
0pauses in Lightning mode
~85%prompt tokens served from cache
$0markup on your AI spend
Engineered for speed

Every second between prompt and done, attacked.

Speed isn't one feature — it's the sum of removing every wait state in the pipeline.

Lightning mode

One toggle in the composer: plans auto-approve, tools auto-run, and the planner skips optional stages. Zero human pauses from prompt to result — for when you trust the task.

Parallel tool batches

When the model requests several read-only calls at once — file reads, searches, git status, MCP queries — they execute concurrently. A batch of five reads costs one round-trip, not five.

Dependency-aware parallel implementers

The plan is split into independent groups using the project's import graph, then implemented by a dynamic fleet of concurrent agents (the graph decides how many — up to 8), each with strictly disjoint file ownership. Competitors parallelize on faith and stay conservative; a proven graph lets us go wider without losing quality.

Three-layer caching

Exact-match and semantic answer caches plus provider-side prompt caching. Repeated context is billed at a fraction of the price — and answered near-instantly.

Scout pyramid

Frontier models never read raw files. A deterministic MCP brief (hybrid search + import graph + file skeletons, zero LLM calls), regex hints from your prompt, and up to three cheap parallel scouts feed the planner conclusions — cutting frontier input cost up to 8x.

Pipelined reviews (quality floor)

Each parallel group is reviewed the moment it finishes — while siblings are still implementing — with an in-place fix cycle. The final full review stays as the unchanged quality gate, and usually passes first try.

The timer is the product

Every task shows a live stopwatch and finishes with its wall-clock time. We benchmark against the fastest tools on the market — and publish the numbers.

ProjectMind MCP: the engine that asks the graph, not the filesystem.

A semantic index of your codebase — symbols, imports, project memory — kept fresh after every change. It's the brain behind Peregrine — and why the planner doesn't burn a million tokens re-reading your repo, and why parallel implementation is safe at all.

01 Semantic retrieval

The planner queries the index for relevant code instead of reading files one by one. Grounded plans, a fraction of the tokens.

02 Dependency graph

Import relations decide which plan steps can safely run in parallel — the scheduling signal competitors simply don't have.

03 Impact analysis

Before review, the agent checks what depends on every touched file and which tests to run — regressions caught before they exist.

04 Project memory

Conventions, decisions, and onboarding notes persist across sessions. Changed files are re-indexed mid-task, so context never goes stale.

Your models, your keys

Frontier brains where it matters. Cheap tokens where it doesn't.

Route each pipeline stage to a different provider: a frontier model for planning and review, a fast cheap model for implementation. Model lists are pulled live from provider APIs — when a new model ships, it's in the picker the same day, no update needed.

Per-stage routing

Planner on Claude, implementer on DeepSeek, reviewer on GPT — mix Anthropic, OpenAI, Google, and any OpenAI-compatible endpoint in one pipeline.

Local models, zero keys

One-click presets for Ollama and LM Studio — point at localhost and run fully offline, no API key required. OpenRouter preset gives you hundreds of models with a single key.

BYOK, no middleman

Your keys are stored locally and calls go straight to the provider. We never proxy your traffic and never take a cut of your usage.

Numbers from real sessions

Collected anonymously from active installations — no code content, no prompts, only aggregated counters.

of input tokens from cache
provider prompt caching — paid at ~10% of price
code approval rate
changes accepted by developers
MCP tool calls
index queries, graph lookups, analysis
agent turns completed
across all active installations
output tokens generated
code, plans, reviews written by agents
Pricing

All the speed is free. The fleet is Pro.

Bring your own keys — we never mark up your AI spend. You pay only for orchestration scale.

Free — forever

Everything that makes a single task fast: Lightning mode, the scout pyramid, parallel implementers, pipelined reviews, stage routing, all providers and local models. Plus 2 parallel tasks with the file-claim registry and git queue.

Fleet — Pro, coming soon

3–10 parallel tasks and a dependency board between them: queue a whole feature list, tasks start the moment their dependencies land. Early adopters get Pro free.

Install in seconds

Desktop app for Windows or a VS Code extension. No account, no license key, no server setup — auto-updates included.

1

Grab the installer: Windows .exe, Linux .AppImage/.rpm, or the .vsix for VS Code

2

Add an AI key in Settings — or pick the Ollama preset and use no key at all

3

Drop an idea into Ideation and watch it drawn — or type a task, flip on Lightning, and watch the timer

Download for Windows Linux .AppImage Fedora .rpm Download .vsix (VS Code)

Latest release · Windows 10/11 · Linux (AppImage auto-updates; rpm for Fedora) · VS Code 1.85+