Rescriptum

Native agent IDE

The native IDE for multi-round agent work.

Token treasury, project context ledger, and copilot mission control — built into a desktop editor, not a web wrapper. BYOK via OpenRouter.

Tauri desktop·Windows Beta · Releasing Next Week·Ask · Agent · Plan · Build

Cloud dashboard
18.4K · $0.08
routes.ts
Agent
routes.ts
import { rateLimit } from "../middleware/rateLimit";
 
export async function handleRequest(req: Request) {
const limited = await rateLimit(req, { windowMs: 60_000, max: 100 });
if (!limited.ok) {
return new Response("Too many requests", { status: 429 });
}
 
const session = await authenticate(req);
if (!session) {
return new Response("Unauthorized", { status: 401 });
}
 
return routeHandler(req, session);
}
patch_file · src/api/routes.ts

Interactive preview — explore sidebar, files, and chat like the desktop app.

Architecture

Three pillars under the hood

Rescriptum isn't a chat box on Monaco. It's a native agent engine with control, memory, and cost discipline baked in.

01

You approve the blast radius

Control

Dangerous tools need approval unless you opt in. Terminal hard-deny rules block catastrophic commands. Git worktrees isolate risky agent work. Per-tool permission presets: strict, balanced, or permissive.

  • Approval gates + hard-deny shell policy
  • Worktree isolation for agent branches
  • Copilot + Agents mission control

02

Context that compounds

Intelligence

Continual harness memory, project context ledger, knowledge graph, and smart reads keep agents sharp across editor and copilot sessions.

  • Project context + harness memory
  • Smart read & symbol lookup
  • Enhance prompt before agent sees it

03

BYOK with real visibility

Economics

300+ models via OpenRouter. Token treasury tracks every call. Prompt caching and auto-routing cut waste. You pay providers directly, with no inference markup.

  • Live treasury & budget caps
  • Cache-aware model routing
  • 300+ models, switch mid-chat
BYOK OpenRouter — inference billed by the provider, not Rescriptum.

Workflow

Four modes, one engine

Ask for quick guidance. Plan without risk. Build with intent. Agent when you want the full loop.

Agent mode

Full autonomy with guardrails

Read, write, patch, run commands, MCP tools, and verify. Default mode for fixes, features, and exploration.

TOOLS this round
read_filepatch_filerun_commandget_diagnosticsverify+ MCP tools

Agent observability

Edit, verify, log. Recovery when tools fail.

Verify runs after edits and surfaces in rollout logs. The editor path prioritizes flow — failed gates become advisory notes on turn_complete. Copilot workers can run with stricter autonomous gates.

Happy path
patch → diagnostics → test → turn_complete
round 1…
rollout.jsonl · debounced verifyclaude-sonnet · $0.11
Loop recovery
patch fails → guardrail → write → verify
round 1…
rollout.jsonl · debounced verifyclaude-sonnet · $0.11

Verify gates

LSP diagnostics, scoped typecheck, and build/test after edits. Debounced so rapid patches don't spam CI.

Turn completion

Model signals done via turn_complete. Gate results attach as notes in editor mode; workers can enforce harder stops.

Rollout logs

Full JSONL trace per turn: every tool call, cost, and round for debugging and audit.

Cloud dashboard

Usage synced from the IDE

Sign in with the same account in the desktop app. Costs, models, and workspace activity update after agent turns — typically within ~60 seconds.

  • Token and cost breakdown by model
  • Per-workspace activity from usage logs
  • Device heartbeat when signed into desktop
Open dashboard
Example layoutlive when signed in
Tokens
2.4M
illustrative
Cost
$4.12
illustrative
This week
Model mix
deepseek-v4-flash 42%claude-sonnet 28%gpt-5.5 18%others 12%

Trust & safety

Serious agents need serious guardrails

Autonomy without oversight is how repos get deleted. Rescriptum layers policy, approvals, and isolation so you can delegate real work.

Terminal security

Hard-deny catastrophic commands (rm -rf /, $HOME, .). Chained shell parsing. Shell substitution blocked on approval-protected ops.

Tool permission presets

Strict, balanced, or permissive defaults in Settings. MCP tools, git push, and deletes can always require approval.

Worktree sandboxing

Agent edits on isolated branches. Review diffs, run verify gates, merge or discard. Main stays safe.

Transparent rollout logs

Every tool call, round, and cost in JSONL. Debug agent behavior without guessing what the model did.

Use cases

Built for real repo work

Multi-round agent sessions on production codebases, not autocomplete demos.

01
Debug

Fix failing tests without a 27-round spiral

prompt › Tests fail on auth middleware. Find root cause and fix.

result › 3 rounds · patch + diagnostics · tests pass · turn_complete

02
UI

Redesign a dashboard with live preview probes

prompt › Make the analytics page responsive. Match the design system.

result › worktree branch · UI verify · Tailwind density check

03
Scaffold

Ship a feature from a Plan → Build workflow

prompt › Plan a webhook ingestion API, then implement after I approve.

result › Plan mode recon · Build mode · verify gates on each file

04
Isolation

Risky refactors in git worktrees

prompt › Migrate the terminal backend to PTY. Don't touch main.

result › agent worktree · review diff · merge or discard

MCP + PlaywrightHarness memoryUI verify probes

Platform

What ships in Rescriptum

Native agent infrastructure for serious repo work — not a chat overlay on an editor.

Economics

  • —BYOK OpenRouter — no inference markup
  • —Live token treasury and budget caps
  • —Cache-aware routing and session stickiness

Control

  • —Hard-deny shell policy + approval gates
  • —Git worktree isolation (auto or explicit)
  • —Strict / balanced / permissive tool presets

Context

  • —Project context ledger across sessions
  • —Harness memory + knowledge graph
  • —Smart read, enhance prompt, MCP tools

Observability

  • —Rollout JSONL audit logs per turn
  • —Verify gates after edits (advisory in editor)
  • —Cloud usage sync when signed in
0+
Models
OpenRouter BYOK
0
Chat modes
Ask, Agent, Plan, Build
Native desktop
Tauri + Monaco
Not a browser wrapper

Pricing

Simple, transparent pricing

Start free with your own key. Pro adds cloud features when billing ships — the IDE stays fully usable on Free.

Beta

Free

$0/mo
  • Full IDE — all agent modes
  • BYOK via OpenRouter
  • Local treasury & rollout logs
  • Cloud usage dashboard when signed in
Coming soon

Pro

$19/mo
  • Extended cloud analytics
  • Priority model routing
  • Cross-device preferences sync
  • Email support
Notify me

Enterprise

Custom
  • Self-hosted deployment
  • Custom model endpoints
  • SSO & team management
  • Dedicated support & SLA
Contact sales

Download

Start with the desktop IDE

Rescriptum for Windows is currently in private beta, with public release launching next week. Request access below to receive an early build or release invitation.

Public release next week·Cloud dashboard

Native desktop · Multi-round agents · Monaco editor