scritty is live · free · last verified 30 Oct 2026 · viability monitor
scritty
A terminal emulator that captures every AI coding agent turn you run inside it and makes that memory searchable across providers.
If you regularly run two or more AI coding CLIs on one codebase, Scritty addresses a problem nothing else quite does: the decision your Claude session made is invisible to Codex, and this is the cleanest fix we have seen. The rules engine that injects prompt.toml, your enforcement rules and each vendor's native rule file on every send is the sleeper feature, and the token-secured browser/PWA view with an explicit Tailscale Funnel opt-in is the part most competitors skip. Skip it if you work in one agent, or if you live in an IDE and never open a terminal. Compare it against staying on tmux plus a notes file if your sessions are short and self-contained.
Verdict generatedVerified 5h ago · liveness 61/100 · cite: rightaichoice.com/tools/scritty
- Developers running Claude Code, Codex, Copilot and others on the same codebase
- Teams that lose architectural decisions at the boundary between AI tools
- Anyone re-pasting the same standing instructions into every agent
- Developers who want a searchable record of AI work that stays on their machine
- Developers who use a single AI agent and have no cross-session memory problem
- Users who work primarily in IDE-side or browser AI chat rather than the terminal
- macOS users unwilling to launch Scritty as a separate terminal on every session
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Scritty if you run a single AI coding CLI and never need to recall what a different agent decided in an earlier session.
Phone pairing needs an explicit Tailscale Funnel opt-in on top of TCP 3000, so budget time for the Funnel origin and a trusted cert rather than expecting a zero-config mirror.
Scritty is a single free download for macOS (Apple Silicon), Windows and Linux, which puts it below paid agent-memory subscriptions and far below per-seat team memory platforms — the trade is that you supply your own machine, your own vector backend and your own Tailscale setup for remote access.
In short
scritty — A terminal emulator that captures every AI coding agent turn you run inside it and makes that memory searchable across providers. Best for Developers running Claude Code, Codex, Copilot and others on the same codebase, Teams that lose architectural decisions at the boundary between AI tools, Anyone re-pasting the same standing instructions into every agent. Free to use.
What is scritty?
scritty — A terminal emulator that captures every AI coding agent turn you run inside it and makes that memory searchable across providers. Best for Developers running Claude Code, Codex, Copilot and others on the same codebase, Teams that lose architectural decisions at the boundary between AI tools, Anyone re-pasting the same standing instructions into every agent. Free to use.
Scritty is a terminal emulator built for developers who run more than one AI coding CLI on the same codebase. Run Claude Code, OpenAI Codex CLI, GitHub Copilot CLI, Gemini CLI, Antigravity CLI, xAI Grok CLI or a local Ollama model inside Scritty and it detects the provider, captures each turn it can attribute — the prompt submitted and the output the pane wrote afterwards — tags it with the provider and indexes it into one corpus stored on your machine. The premise is that the terminal is the one boundary every CLI agent crosses, so that is where the memory gets written. Search is hybrid, mixing vector and keyword retrieval, and it is reachable three ways: Ctrl+Shift+M opens a side panel inside the terminal, the CLI answers the same queries, and your own agents can hit the corpus over MCP. The outbound side is unusual. Before a message reaches the agent, Scritty folds in your standing instructions from prompt.toml (Ctrl+Shift+R), hard enforcement rules (Ctrl+Shift+E), the current goal state to prevent drift (Ctrl+Shift+G) and the provider's native rule file (.claude/CLAUDE.md, AGENTS.md, copilot-instructions.md, GEMINI.md). Write the rules once and every agent gets them, each section toggled live from the terminal. The desktop also embeds a token-secured web server, so the same PTY and memory panel open byte-for-byte in a browser and can be installed as a PWA. Phone pairing runs through an explicit Tailscale Funnel opt-in; TLS is always on, bearer tokens are stored as hashes only, a CIDR allowlist sits in front, and non-loopback clients get a 403 until you allow them. Pluggable vector backends cover qdrant, pgvector, chroma and weaviate. On Linux you can set Scritty as the system default terminal so every shell your desktop or editor opens is already capturing; on Windows you register it as your per-user Default Terminal Application. It suits developers and small teams already running two or more agents on one project.
Key features
- Terminal emulator with auto-capture of prompts and responses from supported AI coding CLIs
- Agent detection and turn attribution for Claude Code, OpenAI Codex CLI, GitHub Copilot CLI, Gemini CLI, Antigravity CLI, xAI Grok CLI and local Ollama
- Hybrid vector + keyword search across all agents and sessions (Ctrl+Shift+M)
- MCP server so your agents can query the shared corpus directly
- CLI memory search and session management
- Rules engine injecting prompt.toml standing instructions on every send (Ctrl+Shift+R)
- Hard enforcement rules injected on every message (Ctrl+Shift+E)
- Goal state kept in front of the agent to prevent drift (Ctrl+Shift+G)
- Reads native vendor rule files: CLAUDE.md, AGENTS.md, copilot-instructions.md, GEMINI.md
- Live per-section toggles for each rules layer from inside the terminal
- Browser sync of the same PTY and memory panel via token-secured embedded web server
- Installable PWA with a standalone app window in Chrome or Edge
- Phone pairing via QR code over an explicit Tailscale Funnel opt-in, with reconnect
- TLS always on, bearer tokens stored as hashes, CIDR allowlist, loopback-only by default
- Pluggable vector backends: qdrant, pgvector, chroma, weaviate
Behind the verdict
The core insight is sound and narrow, which is why it works. Every AI coding CLI — Claude Code, Codex, Copilot, Gemini, Antigravity, Grok — writes its own history in its own format in its own silo. Scritty sits at the terminal, which is the one layer all of them cross, detects which provider is running, and records the prompt and the output the pane wrote afterwards into a single local corpus. That is the whole trick, and it means adding a new agent to the capture set is a provider-detection problem rather than an integration-per-vendor problem. Retrieval is hybrid vector plus keyword, with three front doors: the Ctrl+Shift+M side panel, the CLI, and an MCP server your own agents can query. The MCP route is the one that changes behaviour — an agent that can ask "what did we decide about auth two weeks ago?" stops re-solving it. Pluggable backends (qdrant, pgvector, chroma, weaviate) mean you are not locked into one embedding store. The rules engine is where Scritty does something its competitors do not. prompt.toml holds standing instructions, a separate layer holds hard enforcement rules, a third holds goal state to keep the agent from drifting, and Scritty also reads the vendor's own rule file. All four are injected on every send, each toggled live with a keystroke. If you have been maintaining parallel CLAUDE.md, AGENTS.md and copilot-instructions.md files that drift apart, this is the consolidation. Where it needs care: Scritty is a terminal you have to adopt. On Linux it can be the system default terminal so editors and file managers open into it already capturing; on Windows per-user Default Terminal registration exists. macOS has no such hook, so you launch Scritty deliberately and run agents inside it — a real friction for Mac-based teams. Remote access is deliberately un-zero-configuration: the web listener is loopback-only by default, every other source gets a 403, and phone pairing requires an explicit Tailscale Funnel opt-in, a trusted origin and a bearer token whose hash alone is stored. That is the right default for a corpus of your source code and architecture decisions, but it is setup work. Early-access status means the surface is still moving. The honest comparison set is tmux plus a notes file (free, no retrieval, no rules injection) and vendor-native session history (searchable within one tool, siloed by design). Scritty wins when the cross-agent question is real and recurring. It adds little if you run one agent, or if you never leave the IDE.
Models under the hood
as of 2026-10-24
Limitations
- Scritty is in early access, so the feature set and stability are still evolving.
- Capture is only as good as attribution: it records each turn it can attribute, so a provider that changes its CLI output format can degrade capture until Scritty catches up.
- On Linux it can be the system default terminal so every shell your desktop or editor opens is already capturing, and on Windows per-user Default Terminal registration exists. macOS has no system default-terminal hook at all: you launch Scritty and run agents inside it.
- Captures stay on your machine, and remote access takes an explicit Tailscale Funnel opt-in plus a trusted origin, so phone pairing is not a zero-configuration mirror.
- Non-loopback clients receive a 403 until you widen SCRITTY_WEB_BIND_ALLOW or pair a token.
as of 2026-10-30
Researching scritty? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Is scritty still active?
scritty is live · free · last verified 30 Oct 2026 · viability monitor.
Viability score
How well maintained and how widely used is scritty? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →What do users say about scritty?
What people actually say about scritty — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
ResearchedDiscussed by users across 3 sources (Hacker News, YouTube, Product Hunt) · researched Sep 14, 2026.
Weighted by the 63 posts each of 3 sources contributed.
- +Passive capture inside the terminal means no wrapper or workflow change to adopt it
- +One shared memory across Claude Code, Codex, Copilot, Antigravity, and Ollama ends context re-pasting
- +Local-first storage plus at-rest encryption keeps the corpus on your machine
- +MCP server closes the loop, letting agents query past context programmatically
- +Prompt.toml rules engine counters rule-relevance decay in long sessions
- −No clear mechanism to mark stale or wrong memories as superseded
- −Unconfirmed whether API keys in raw terminal output get redacted at capture
- −Cross-model memory legibility is unproven when different agents read each other's notes
- −How agents are hooked — wrapper versus passive observation — is still unclear
- −Early access with no independent GitHub or Reddit reliability signal
- • Vector backend choice (qdrant, pgvector, chroma, weaviate) may carry its own hosting or storage bill
- • Optional cloud sync could become a paid tier once early access ends
- • Team tenancy and admin override features hint at a future enterprise price point
How we verified this
Every listed fact about scritty— pricing, features, links, status — is re-read from the vendor's own pages by our verification pipeline and each change is logged with a date; the last pass ran on 30 Oct 2026. Pages mix model-assisted synthesis with human review; see how we verify.
Verification history
We have re-verified scritty 9 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — changed summary, who should skip it, setup time, features, who it suits, hidden costs, use cases, switching in, limitations, pricing fit, our verdict, description, switching out
- — changed SSL certificate
- — changed SSL certificate
- — changed description, who it suits, use cases, setup time, our verdict, summary, features, hidden costs, limitations, switching in, switching out, who should skip it, pricing tiers, pricing fit
- — changed pricing page, platforms
- — changed who should skip it, our verdict, summary, who it suits, pricing tiers, integrations, description, features
Showing the 6 most recent of 9 verification passes.
Free to cite with attribution — this page re-verifies continuously.
How much does scritty cost in 2026?
scritty: Free to use. Pricing last verified 30 Oct 2026.
Plans compared
Every published scritty tier, its price as the vendor publishes it, who it fits, and what it adds over the tier below. The 12-month cost projection follows.
| Plan | Price | Billing | Who it's for |
|---|---|---|---|
| Free | Free | — | Developers already running two or more AI coding CLIs on one codebase who want a searchable, locally stored record across all of them.Adds: Starting tier and the only published one: full terminal capture, hybrid search, rules injection, MCP and CLI access, browser/PWA sync and phone pairing over Tailscale Funnel. |
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Where the pricing makes sense
The company stage and team size where scritty's pricing actually pencils out — and where peers do it cheaper.
Scritty is a single free download for macOS (Apple Silicon), Windows and Linux, which puts it below paid agent-memory subscriptions and far below per-seat team memory platforms — the trade is that you supply your own machine, your own vector backend and your own Tailscale setup for remote access.
Has scritty's pricing changed?
No plan price for scritty has moved since we began logging in Aug 2026; the pricing page is re-read on every verification pass.
Who should pick scritty?
Pick scritty if you are Developers running Claude Code, Codex, Copilot and others on the same codebase, Teams that lose architectural decisions at the boundary between AI tools, Anyone re-pasting the same standing instructions into every agent, Developers who want a searchable record of AI work that stays on their machine.
Look elsewhere if you are Developers who use a single AI agent and have no cross-session memory problem, Users who work primarily in IDE-side or browser AI chat rather than the terminal, macOS users unwilling to launch Scritty as a separate terminal on every session, Teams that need an on-premise enterprise deployment rather than local installs.
Skip it if Skip Scritty if you run a single AI coding CLI and never need to recall what a different agent decided in an earlier session.
GeneratedReal-world workflow fit
Concrete scenarios for the personas scritty actually fits — and what changes day-one when you adopt it.
Claude Code settles the auth middleware design on Monday; Tuesday you start a Codex session on the same repo and hit Ctrl+Shift+M to search the corpus before re-deciding.
Outcome: The Codex session starts from the Monday decision instead of re-deriving it, and the prompt/output pair stays searchable in your local index.
You write the house style once in prompt.toml plus enforcement rules, and Scritty injects them along with each vendor's native rule file on every send, with per-section toggles from the terminal.
Outcome: CLAUDE.md, AGENTS.md and copilot-instructions.md stop drifting apart because the standing instructions reach every agent from one place.
You start a refactor in the desktop window, opt into a Tailscale Funnel, run scritty pair and scan the QR on your phone, then drive the same PTY from the train.
Outcome: Keystrokes reach the same session as the desktop, and searching the corpus from the phone returns the same results as the local memory panel.
Use cases
- Store coding conventions and preferences once so Claude, Codex and Copilot all follow the same style.
- Save a bug fix to memory so a later session in a different agent doesn't reintroduce the issue.
- Keep architecture decisions queryable by every agent instead of re-explaining them per tool.
- Ask "what did we decide about auth two weeks ago?" from the side panel or the CLI.
- Have an agent query the corpus over MCP for past solutions instead of re-solving a known problem.
- Capture test patterns and library usage automatically so a new agent starts with context.
- Monitor or drive a long-running agent from a browser tab or a paired phone.
- Run offline on a plane with a local Ollama model and still search the same corpus.
Setup time & first value
How long it actually takes to get something useful out of scritty — broken out by persona, not the marketing-page minute.
Desktop capture is minutes: install Scritty, run an agent inside it, and Ctrl+Shift+M has something to search. Browser sync is immediate on loopback. Phone pairing takes longer — a Tailscale Funnel, a trusted cert and a scritty pair run before the QR works. Setting it as the Linux default terminal takes effect for every new shell afterwards.
Switching to or from scritty
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From tmux plus a notes file: keep the tmux muscle memory, but let Scritty capture turns automatically and search them across agents instead of grepping markdown.
- →From per-vendor rule files only: move shared instructions into prompt.toml and leave the vendor files to carry only tool-specific rules.
- →From vendor-native session history: run the same CLIs inside Scritty instead of standalone so the turns land in one corpus.
- ↗To tmux plus a notes file: you keep a terminal you configure yourself but lose cross-agent capture, hybrid search and rules injection.
- ↗To vendor-native session history: nothing to migrate technically, but each agent's log is siloed again with no cross-agent retrieval.
- ↗To a hosted team memory service: your local corpus would need exporting, and captures would leave the machine that currently holds them.
Integrations
Resources & guides
Tutorials & Learning
YouTube returned 6 videos for “scritty”, and we withheld 6: 6 could not be judged, because “scritty” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about scritty.
Official links
What are the best scritty alternatives?
The closest alternatives to scritty we list are Wato, FlowLens and Aghub. Ranked by direct product-type match, not category overlap.
| Alternative | Pricing | Starts at | Last verified |
|---|---|---|---|
| Wato | Contact Sales | Custom pricing | — |
| FlowLens | Freemium | Pricing not published | — |
| Aghub | Free | Free | — |
Alternatives to scritty
View all
Tools that pair well with scritty
Common stack mates teams adopt alongside scritty, with the specific reason each pairing earns its keep.
Wato
Wato is a shared AI workspace that connects every team's coding and chat agents through one MCP server for memory, tools, and cloud
FlowLens
Chrome extension and MCP server that captures the full browser context of a bug and hands it to your AI coding agent as one link.
Aghub
Desktop hub that deploys one MCP server config and shared .skill packages across 23 AI coding agents
scritty vs its closest rival
scritty's closest head-to-head is with Cognition Ai — If you're an enterprise team needing an autonomous agent that plans, codes, and ships production code end-to-end, Cognition AI's Devin is the clear choice—it's backed by a $10M guarantee and Fortune 500 deployments. For individual developers or small teams using multiple AI coding assistants who want persistent, shared memory across sessions, Scritty is a free, lightweight solution that solves context repetition. They address completely different problems: autonomous engineering vs. context sharing. Verdict generated
Head-to-head comparisons
Scritty vs Cognition Ai
If you're an enterprise team needing an autonomous agent that plans, codes, and ships production code end-to-end, Cognition AI's Devin is the clear choice—it's backed by a $10M guarantee and Fortune 500 deployments. For individual developers or small teams using multiple AI coding assistants who want persistent, shared memory across sessions, Scritty is a free, lightweight solution that solves context repetition. They address completely different problems: autonomous engineering vs. context sharing.
GeneratedScritty vs Bito
For engineering teams managing complex multi-repo architectures with AI coding agents, Bito's knowledge graph and automated scoping are indispensable. Scritty is excellent for individual developers or small teams wanting consistent context across agents, but lacks the architectural depth and enterprise features Bito offers. Choose Bito for system-wide context, choose Scritty for lightweight persistent memory.
GeneratedFrequently asked questions
Best-of guides
Topics
Used scritty? Help shape our editorial sentiment research.