Migration snapshot
Shut downPlatform
an Autocode
Autocode (formerly Standard Library, operated by Polybit Inc.) shut down April 26, 2024. Users were emailed April 3, 2024 with 23 days' notice. Crunchbase lists the company as permanently closed. The founder joined OpenAI. The GitHub CLI repo (github.com/acode/cli) remains online but the runtime that executed code is offline.
Typical timeline
4–8 weeks
Typical cost
$13K–$25K (agency, fixed)
Why teams leave an Autocode
Autocode is gone. Anyone with Autocode webhook URLs in production services, scheduled functions feeding live systems, or API endpoints built on Autocode has broken dependencies that have been failing silently since April 26, 2024.
Platform shut down April 26, 2024
Polybit Inc. announced closure April 3, 2024 with 23 days' notice. The hosted service went offline April 26, 2024; all functions stopped executing. Crunchbase lists the company as permanently closed.
No self-host runtime was ever released
Autocode's runtime (FunctionScript, Node.js) ran exclusively in Autocode's cloud. No self-hosted version or OSS release was ever made available. There is no path to run your functions outside Autocode's servers.
GitHub CLI repo exists but is non-functional
github.com/acode/cli remains online and can pull code you wrote if you used the CLI. However, the execution environment it pointed to is gone — the repo is a code-recovery tool only.
15+ months of broken integrations
Any Autocode-hosted webhook, scheduled function, or API endpoint has been returning errors since April 26, 2024. Downstream systems that relied on these have been degraded for over 15 months.
Webhook URLs are dead
Autocode provided hosted webhook URLs (*.stdlib.com or *.autocode.com). These no longer respond. External services (Slack, GitHub, Stripe, Airtable) configured with these URLs have been failing silently.
What can you actually take with you?
Code recovery depends entirely on whether you used the Autocode CLI. Browser-only workflows with no CLI sync may be unrecoverable. External service data is in those services, accessible directly.
| Asset | Can you export it? | How | Notes |
|---|---|---|---|
| Data | Partial | If Autocode functions read/wrote to external services (Airtable, Google Sheets, Slack, GitHub), that data is in those systems and accessible directly. | Autocode did not provide a persistent database. Any in-function state not written to an external service is unrecoverable. |
| Code | Partial | If you used the Autocode CLI (`lib` command) and had local code synced, your Node.js function code exists on your machine. Pull from github.com/acode/cli if you used it. FunctionScript function definitions (parameters, return types) are standard-ish JSON and can be read. | Code written exclusively in the browser IDE with no CLI sync and no local backup may be unrecoverable. Check browser cache and IndexedDB before concluding code is gone. |
| Design/UI | No | Not applicable — Autocode was a serverless API and integration platform with no app UI. | N/A |
| Logic/Workflows | Partial | FunctionScript function definitions (parameters, return types, authentication tokens) can be translated to standard Node.js/TypeScript if recovered from CLI. The logic itself is portable if you have the source. | Autocode's standard library of 1,000+ API connectors had managed OAuth; those tokens are gone and must be re-authorized fresh. |
| Users & Auth | No | Autocode managed workspace auth. User accounts are disabled. | N/A — rebuild access control from scratch; no user list is recoverable. |
Swipe the table sideways to see the full breakdown.
Where each piece moves in code
The target stack is Next.js + Supabase/PostgreSQL on Vercel, replacing Autocode's managed FunctionScript runtime with standard Node.js/TypeScript serverless functions.
an Autocode
Function (Node.js/FunctionScript)
In code
Next.js Route Handler or Vercel Serverless Function
Autocode functions are Node.js; port directly. Replace `context.params` pattern with standard Next.js `request` parameter.
an Autocode
Webhook endpoint (*.stdlib.com or *.autocode.com)
In code
Next.js Route Handler URL on Vercel
Update all external services (Slack, GitHub, Stripe) with new webhook URLs immediately after deploy — this is the first action.
an Autocode
Scheduled function
In code
Vercel Cron Jobs or Supabase pg_cron
Reconstruct schedule from Autocode dashboard screenshots or documentation if available. Add failure alerting from day one.
an Autocode
API connector (managed OAuth)
In code
Supabase Vault or server environment variables + individual SDK
Re-authorize each connected app (Slack, GitHub, Airtable) via its own OAuth flow. Managed tokens are gone.
an Autocode
Data passing between functions
In code
Next.js Server Actions with Supabase as shared state
Replace implicit function chaining with explicit state management using Supabase tables.
an Autocode
Team auth / user tokens
In code
Supabase Auth + Clerk or NextAuth
Rebuild access control from scratch. No existing user list is recoverable from Autocode.
The migration roadmap
Recovery starts with finding the code and auditing broken integrations — not writing new code. The first 48 hours should focus on replacing dead webhook URLs in external services.
Phase 1: Code Recovery & Dead URL Audit (urgent)
Days 1–3 (urgent)- Run `grep -r "stdlib.com\|autocode.com"` across your entire codebase, CI config, and documentation to find all dead webhook URLs
- Check every external service (Slack, GitHub, Stripe, Airtable) for webhook configurations pointing at Autocode URLs — update or disable immediately
- Pull any available code from github.com/acode/cli if the CLI was used
- Check browser IndexedDB and cache for any IDE sessions with unsynced function code
- Search team Slack/email history for function code snippets shared informally
Watch out: Dead webhook URLs are causing live failures in production integrations right now. Address these before doing anything else.
Phase 2: Inventory & Gap Assessment
1 week- List every Autocode function that ran on a schedule and identify its downstream consumer (report, alert, database table)
- Identify every OAuth-connected app and its admin — schedule re-authorization sessions
- Triage recovered functions by business impact: which are customer-facing, which are internal monitoring only
- Document the 15-month data gap in any reports or systems that depended on Autocode-scheduled outputs
Phase 3: Rebuild & Re-authorize
2–5 weeks- Set up Next.js on Vercel with Supabase connection; configure Supabase Vault for secrets
- Port recovered Node.js function code to Next.js Route Handlers; replace FunctionScript patterns with standard Next.js request handling
- Re-authorize each OAuth-connected app (Slack, GitHub, Airtable, Stripe) via its own OAuth flow; test each before proceeding
- Rebuild scheduled functions on Vercel Cron; add failure alerting immediately
- Update all external services with new Vercel webhook URLs
Phase 4: Validation & Documentation
1 week- Validate all rebuilt functions produce correct output; test each webhook endpoint with a real trigger
- Document the new stack architecture and function inventory for team onboarding
- Acknowledge the 15-month data gap explicitly in any affected dashboards or reports
- Remove all references to dead Autocode URLs from codebases, CI configs, and runbooks
Three ways to migrate — honestly
Every path has a real trade-off. Here is what each costs, how long it takes, and where it bites.
DIY (with AI tools)
$0–$500 + time
2–6 weeks part-time
Fits
Teams where the CLI was used and clean Node.js code is recoverable. If the scope is a handful of functions and one or two webhook endpoints, this is a straightforward porting exercise.
Risks
OAuth re-authorization for multiple connected apps (Slack, GitHub, Airtable) requires admin access to each service and takes longer than expected. Code not in CLI sync or Git may be permanently unrecoverable.
Freelancer
$3K–$10K
2–6 weeks
Fits
Teams with recovered code and a clear inventory of functions and external service connections. Works best when scope is bounded and does not include complex function chaining or multi-app OAuth flows.
Risks
Freelancers rarely scope the forensic gap-analysis phase adequately. Ensure the contract covers dead URL audit and OAuth re-authorization, not just code porting.
Agency (RapidDev)
Done-for-you$13K–$25K fixed
4–8 weeks
Fits
Teams with unknown scope, significant dead webhook coverage, multiple OAuth-connected apps, or critical downstream systems that have been broken for months.
Risks
Minimal — fixed price covers forensic audit, code recovery assessment, full rebuild, OAuth re-authorization, webhook migration, and cutover. Free scoping call at rapidevelopers.com.
The real risks — and how to defuse them
Code unrecoverable if no CLI sync was used
Mitigation: Functions written in the browser IDE with no CLI sync and no external Git backup may be permanently lost. Check browser cache and IndexedDB before concluding code is gone. Check team Slack/email for function code snippets shared informally. Set a bounded time limit for recovery archaeology.
Dead webhook URLs causing live silent failures
Mitigation: External services (Slack, Stripe, GitHub) are sending events to dead Autocode URLs. Audit every external service integration immediately and update or disable webhook URLs as the first priority — before any other rebuild work.
OAuth re-authorization requiring admin access
Mitigation: Each connected OAuth app needs permissions reviewed and re-granted. Some apps (internal Slack bots, GitHub Apps) require admin access to reconfigure. Identify the admin for each OAuth app early; schedule re-authorization sessions with them — this is often a scheduling bottleneck, not a technical one.
Business logic gaps from incomplete recovery
Mitigation: Functions with complex branching logic not captured in comments or tests are hardest to reconstruct. Interview the developer who originally built each function. Set a maximum effort limit per function for reconstruction — archaeology is time-bounded.
Assuming the platform will come back
Mitigation: Crunchbase lists Autocode/Polybit as permanently closed; the founder joined OpenAI. Do not wait for recovery. Treat as permanently dead and start the rebuild immediately.
Should you actually migrate?
Migrating is a real project. Sometimes staying is the right call — here is the honest split.
Stay if
- There is no stay option — Autocode shut down permanently on April 26, 2024
- If the only affected workflows are internal-only monitoring scripts with no external dependencies, you may deprioritize rebuild and document the gap until it becomes blocking
- If you recovered clean code from CLI sync and the scope is limited, you may need only 1–2 weeks of porting work rather than a full agency engagement
Migrate if
- Any external webhook URL (*.stdlib.com or *.autocode.com) is registered in a production service (Slack, Stripe, GitHub) — fix immediately
- Any scheduled function was producing data consumed by a live dashboard, report, or alert — the data has a 15-month gap that must be acknowledged and addressed
- You have inherited this codebase and need to understand its full scope before decommissioning or handing it off
Our honest verdict
This is a forensic recovery scenario. Lead with an immediate audit of dead webhook URLs, then prioritize recovery of function code, then rebuild on a standard serverless stack. The sooner broken integrations are replaced, the less manual workaround accumulates in the business.
Do this today: pre-migration checklist
Whatever path you choose, protect yourself first. Work through this before you touch a line of code.
Run `grep -r "stdlib.com\|autocode.com"` in your entire codebase, CI config, and documentation to find all dead webhook URLs
Dead webhook URLs are the most urgent live problem — external services are silently failing right now while you read this.
Check every external service (Slack, GitHub, Stripe, Airtable) for webhook configurations pointing at Autocode URLs — update or disable immediately
Each dead URL in a production service is an active silent failure. Disable or replace before starting any other rebuild work.
Pull any available code from github.com/acode/cli if the CLI was used; check browser IndexedDB and cache for IDE sessions
Recovered code is porting work; lost code is reconstruction from memory. Maximize recovery before giving up on any function.
List every function that ran on a schedule and identify its downstream consumer
Scheduled functions feeding live dashboards or reports have created a 15-month data gap that must be assessed and acknowledged.
Identify every OAuth-connected app and its admin — schedule re-authorization sessions
OAuth re-authorization requires admin access to each external service. Scheduling bottlenecks are the #1 cause of extended rebuild timelines.
Deploy replacement Vercel Functions for any live webhook endpoint within 48 hours of starting — restore data flow before completing the full migration
Every day with a dead webhook is another day of accumulated manual workaround for the team consuming that integration.
Frequently asked questions
Is Autocode shut down or is it still operating?
Autocode is permanently shut down. Polybit Inc. announced closure on April 3, 2024 with 23 days' notice. The service went offline April 26, 2024. All accounts are disabled; prepaid contracts were refunded. Crunchbase lists the company as permanently closed and the founder joined OpenAI. Do not contact Autocode support — there is none.
Can I export my Autocode function code?
Partially, if you used the Autocode CLI. Pull from github.com/acode/cli — it can help extract code you had locally synced. Functions written exclusively in the browser IDE with no CLI sync may be unrecoverable. Check your browser's IndexedDB and cache before concluding code is lost, and search team Slack/email for code snippets shared informally.
What happens to my Autocode webhook URLs?
All *.stdlib.com and *.autocode.com webhook URLs are dead and returning errors. Any external service (Slack, GitHub, Stripe, Airtable) that was configured to call these URLs has been failing silently since April 26, 2024. Audit every external service for these dead webhook URLs immediately and update or disable them — this is the highest-urgency action.
How long does an Autocode rebuild take?
Typically 4–8 weeks. The largest variables are: whether function code was recovered from the CLI, how many OAuth-connected apps need re-authorization, and how many scheduled functions need reconstructed schedules. Teams with clean CLI-recovered code and a simple function count can move faster. The forensic archaeology phase (days 1–3) is mandatory regardless of scope.
What should I use instead of Autocode?
Autocode was a Node.js serverless function host with managed API connectors. The direct equivalent is Vercel Serverless Functions (Next.js Route Handlers) for the execution layer, with Supabase Vault for secrets management and individual SDKs (Slack SDK, GitHub API, Stripe SDK) replacing Autocode's managed OAuth connectors.
What if I can't recover my Autocode function code?
Start from what you can document. Interview the developer who built each function. Audit connected databases for tables that Autocode wrote to — the schemas and recent data rows reveal what the function did. Review any runbooks, README files, or PR descriptions that described the function's behavior. Business-logic archaeology has a bounded return; set a maximum effort per function before moving to reconstruction from scratch.
How much does an Autocode migration cost?
If CLI-recovered code is clean and scope is limited, a developer can port to Vercel Functions in 1–2 weeks at near-zero cost. For teams with unknown scope, multiple dead webhook URLs in production, and multiple OAuth apps to re-authorize: freelancer work runs $3K–$10K; a fixed-price agency engagement runs $13K–$25K over 4–8 weeks. RapidDev offers a free scoping call to assess recovered code and identify the full scope of broken dependencies at rapidevelopers.com.
How do I know which of my systems depended on Autocode?
Search everywhere: run `grep -r "stdlib.com\|autocode.com"` across your codebase, CI config files, documentation, and runbooks. Check every external service's webhook configuration panel (Slack App settings, GitHub repository webhooks, Stripe webhook endpoints, Airtable automations). Look for any scheduled reports or data-freshness monitoring that has been stale since April 2024.
We migrate no-code apps to production code
- Fixed price — $13K–$25K (agency, fixed)
- No data loss, no downtime
- You own 100% of the code
30-min call. Quote within 48 hours.