On this page
Flowise did not switch your instance off. It stopped being maintained. That is a slower problem than a shutdown, and a more confusing one, because everything still works on the morning you read about it.
Three dates matter. On 29 July 2026 the team announced the wind-down and froze the code. On 10 August the GitHub repository was archived, issues and pull requests were locked, and the npm packages and Docker images were marked deprecated. On 31 August the core team's presence in Discord and GitHub concluded. I read the vendor's own notice at flowiseai.com/sunset on 13 September 2026, and it is signed "Henry & Flowise Team".
Nothing in that notice turns off what you are running. The Apache 2.0 source is still on GitHub, the notice tells you the code is yours to keep building on, and a container that works today will work next month.
What stops is the patching. Two vulnerabilities were published after the final release, and one of them scores 9.9.
So this post does two jobs. First it answers the question most alternatives lists skip, which is whether to fork. Then it lists seven places to land, with my own product last and the reasons you would not pick it stated in the same paragraph as the reasons you would.
The decision in nine lines
- Nothing switched off. Self-hosted Flowise keeps running and the repo stays public under Apache 2.0.
- What ended is maintenance. The final release was
flowise@3.1.4on 29 July 2026, the repo was archived on 10 August, and no release has followed. - The security cost is concrete. CVE-2026-67622 (CVSS 3.1 base 9.9, cross-workspace credential access) and CVE-2026-67620 (SSRF to cloud metadata endpoints) both affect versions through 3.1.4, and both were published after it.
- Fork if Flowise is infrastructure: customised, load-bearing, and maintained by someone who reads TypeScript. A fork inherits both CVEs on day one.
- Closest visual substitute: Langflow, MIT licensed and free to self-host.
- Closest self-hostable platform with a company behind it: n8n, with a licence caveat, or Activepieces if you want their cloud at the same price as your own box.
- If the flows were really RAG apps: Dify.
- The direction the sunset notice itself points at: LangGraph, in code, if an engineer owns this.
- No servers at all: Zapier Agents, or Gravity if you want the task done and no canvas to build on.

What actually happened, and what did not
Two different things are easy to blur here: an open-source project being archived, and your instance stopping. Only one of them happened.
| Date | Milestone | What actually happens |
|---|---|---|
| 29 July 2026 | Announcement and code freeze | Active feature development ceases immediately. No new pull requests are reviewed or accepted. |
| 10 August 2026 | Repository archival | The GitHub repo moves to public archive. Issues and pull requests are locked. npm packages and Docker images are marked deprecated. |
| 31 August 2026 | End of life | The core team's presence in Discord and GitHub concludes. Community moderation is handed off where applicable. |
Milestones as published in the Flowise sunset notice at flowiseai.com/sunset, read 13 September 2026. The GitHub API returns archived: true for FlowiseAI/Flowise, a last push of 13 August 2026 and 55,456 stars.
Read the right-hand column again. None of those rows deletes an account, expires a licence or stops a container. I have written about tools that did exactly that, and this is materially gentler than the way Relay.app wound down, where dated deadlines came attached to the data.
The vendor's stated reason is worth quoting in full, because it explains the shape of this list.
"As AI models become more capable at reasoning, we've noticed that developers are increasingly relying on new coding agents such as Claude Code/OpenClaw to handle complex tasks. The typical rigid workflow low code approach quickly hits the limit when it comes to complexity."
Now the part that decides your week. The final published version is flowise@3.1.4, dated 29 July 2026, the same day as the freeze. CVE-2026-67622 is an insecure direct object reference in the OpenAI Assistants integration: an authenticated attacker supplies an arbitrary credential UUID and reads credentials belonging to other workspaces, enumerates assistant metadata, retrieves file and vector-store listings, and uploads files into somebody else's workspace. It carries a CVSS 3.1 base score of 9.9 and was published on 6 August 2026.
CVE-2026-67620 is the second one. It is a server-side request forgery in the guard in httpSecurity.ts, where the deny list omits the Oracle Cloud metadata endpoint at 192.0.0.192 and the Alibaba Cloud metadata endpoint at 100.100.100.200, so a crafted URL to the fetch-links API can reach cloud metadata services and disclose temporary role credentials.
Both affect every version up to and including the final release. Both were published after it. No release has been issued since. I am not going to predict what does or does not get patched from here, and neither disclosure reflects badly on the team, since both landed roughly a week after the wind-down was already public. The operational fact is enough on its own: if your Flowise is reachable from the internet, that is the thing to deal with before you shortlist a replacement.
One more boundary. The sunset notice says nothing about Flowise Cloud, and flowiseai.com was still serving a Pricing page when I read it on 13 September 2026. Plenty of posts will tell you the hosted product is dead. The vendor has not said that, so I will not either.
Should you fork instead of switching?
Ask this before you open a single pricing page. Dead-tool posts default to a switch-now framing, and for Flowise that default skips a real option, because the licence and the archive make forking a supported path.
The permission is explicit. Apache 2.0 does not expire when a repository is archived, and the sunset notice tells readers the code is theirs to keep building on. The archive locks issues and pull requests. The code stays where it is. You can clone it today, and you can clone it in three years.
Now the bill. Forking is a maintenance contract you sign with yourself, and the first clause is those two CVEs. A fork inherits both. You also inherit model provider SDK churn, which moves every few weeks, plus Node and Docker base-image upgrades forever, plus your own build and publish pipeline, because the npm and Docker channels carry a deprecation warning now.
Here is the rule of thumb I would use. Fork if Flowise is infrastructure: if you have already customised it, if real revenue depends on flows that would take weeks to rebuild, and if somebody on the payroll reads TypeScript and will still be there next year. Do not fork if Flowise was a convenience, which is the far more common case. A convenience is worth rebuilding. Infrastructure is worth owning.
There is a third answer that almost nobody writes down, and for a lot of readers it is the correct one. Freeze it. Pull the instance off the public internet today, put it behind a VPN or an internal network boundary, keep it running for the flows that already work, and build nothing new on it. That buys you months to rebuild at a sane pace. The whole bill is an afternoon of network configuration.
Whatever you choose, do the network work first and the tool selection second. The shortlist below will still be there on Monday.
Must read before you fork: self-hosted AI agents walks through what owning the box actually costs in upgrades, uptime and credential handling. That post is the honest counterweight to "it is free, I will just run it myself".
How I compared them
Same five questions for every tool on this page. They tell you more than a feature grid does, because they ask what you own after the move. The box on the website answers a different question.
- Can you run it yourself, and under what licence? This is the question that sent you here. Apache 2.0 and MIT give you more than "source available" does, and the difference matters the next time a vendor changes direction.
- Who does the work? You sit at a canvas and wire the flow. Or somebody else built the worker already and your part is signing in and approving what it sends.
- What do you own after the move? Somebody still has to apply the next security patch. This question only asks who that is.
- How does the bill grow? Per run, per step, per seat, per credit, or flat. That one line explains most of the surprise invoices in this category.
- How close is the shape to a Flowise chatflow? A visual LLM graph, a general automation canvas and a Python state machine are three different jobs wearing similar words.
What I actually did, stated plainly so you can discount it correctly. On 13 September 2026 I read the Flowise sunset notice, the GitHub API record for the repository, the release history for version 3.1.4, and the published entries for both CVEs. The dollar figures for n8n, Activepieces and Zapier come from those vendors' own pricing pages, read on 11 September 2026 and carried over here with their dates attached.
For Langflow, Dify and LangGraph I report the licence and the cost of running it yourself, because I did not re-read their hosted pricing pages for this post. Check those before you budget. I have not run a bake-off across seven tools, and I am not going to pretend otherwise.
Quick comparison
| Tool | What it is | Who does the work | Starting price (Sep 2026) | Free tier | Best for |
|---|---|---|---|---|---|
| n8n | Source-available automation platform, self-hostable | You build | Free self-hosted; Cloud Starter 20 euros/mo billed annually | Community Edition | A canvas with a company behind it |
| Langflow | MIT-licensed visual builder over LangChain components | You build | Free self-hosted; managed version sold separately | Self-hosting is free | The closest match to Flowise's shape |
| Dify | Open-source LLM app platform with retrieval built in | You build | Free self-hosted; Cloud tiers on their pricing page | Free Sandbox tier on Cloud | Document question-answering and RAG apps |
| Activepieces | Open-source automation builder, their cloud or yours | You build | $16/mo Plus, billed yearly | 100 credits a day | Open source without owning a server |
| LangGraph | MIT-licensed Python library for stateful agents | You write it | Free library; managed platform priced separately | The library is free | Engineers who outgrew the canvas |
| Zapier Agents | Hosted AI agents inside a Zapier account | The agent, you supervise | Professional from $19.99 a month on annual billing | 100 tasks/mo on the free plan | Nothing to host, ever |
| Gravity | AI agent platform, expert-built agents, closed source | The agent. You describe and approve. | $20/mo Autopilot worldwide; in India ₹399 Minipilot or ₹1,999 Autopilot | First agent free | Recurring tasks, nothing to maintain |
n8n, Activepieces and Zapier figures are what those vendors' pricing pages showed on 11 September 2026. Langflow, Dify and LangGraph are shown by licence and self-hosting cost because I did not re-read their managed pricing for this post. Gravity pricing is current as of September 2026. n8n prices in euros; no conversion applied.
1. n8n: a canvas with a company behind it
If what you want is a visual builder running on your own box by tomorrow, this is the shortest move on the page. n8n is self-hostable, the Community Edition costs nothing, and the node graph is close enough to a chatflow that an afternoon gets you productive.
Read the licence before you commit, because this is where a Flowise refugee gets caught. n8n ships under the Sustainable Use License, which the company calls fair-code. You can run it, modify it and use it internally without paying. You cannot sell it as a hosted product. If the Apache 2.0 grant was part of why Flowise was in your stack, moving here is a reduction in rights, and it is easy to miss on a pricing page.
What you buy in exchange is the thing that just failed you. There is a company, a paid cloud, a funded team and a release cadence, so the next security report has somewhere to go.
Why an n8n box costs about what a Flowise box cost
Self-hosting swaps an invoice for a person, and it is the same person you already have. The server, the upgrades and the backups stay yours exactly as they were. The single thing that changes is that upstream is still shipping, so patching is a docker pull instead of a project.
Pros
- Free to self-host, with a maintained upstream
- Per-execution billing on Cloud, whatever the node count
- Large node catalogue and an active community
- Native AI and agent nodes, so LLM flows have a home
Cons
- Source available, not open source, and the difference is real
- You still own the server, the upgrades and the backups
- Automation-shaped, so LLM agent flows need rethinking
- Cloud tiers step up steeply after Pro
- Best for
- A technical team that wants the canvas back on day one and will host it themselves anyway.
- Pricing
- Community Edition free to self-host. Cloud Starter 20 euros a month billed annually for 2,500 executions, Pro 50 euros for 10,000. Read on n8n's pricing page on 11 September 2026.
- Free tier
- The Community Edition is the free tier.
- Verdict
- The safest landing spot if you can live with fair-code. Best n8n alternatives covers the rest of that field, and Gravity vs n8n is the build-versus-buy version of the same question.
2. Langflow: the nearest thing to Flowise's exact shape
Langflow is the substitute if what you liked was the drag-and-drop graph sitting on top of LangChain components. Same idea, same shape on screen, different project.
The licence is MIT, so self-hosting costs you infrastructure and model tokens and nothing else. A managed version is sold separately by DataStax, which is the route to take if the whole point of leaving was to stop running servers.
One caution, and it applies to every open-source project you move to after this week. A repository being busy today is not a promise about next year. Flowise had 55,456 stars when I pulled the GitHub API record on 13 September 2026, and it is archived. Popularity was never the signal. Look at commit frequency over the last ninety days, at how many distinct people have merge rights, and at who is paying them.
The migration is a rebuild even here
Langflow is the closest visual match on this page and it still cannot import a Flowise chatflow. You will be recreating the graph by hand, node by node, reading your exported JSON on one screen and rebuilding on the other.
Pros
- MIT licence, so your rights are at least as good as before
- The visual LLM graph you already know how to think in
- Free to self-host, with a managed option if you want one
- LangChain component model, so the ecosystem is large
Cons
- No Flowise importer, so every flow is a manual rebuild
- Self-hosting puts you back in the same maintenance seat
- Managed pricing sits outside the open-source project
The reason to pick it is continuity. You keep the mental model, your team keeps its vocabulary, and the diagrams in your internal docs still mean something. That is worth more than it sounds when you are also explaining to a client why the thing needs rebuilding at all.
If your instance was three chatflows and an embed, budget an afternoon per flow and start with the one you use daily. Gravity vs Langflow is the comparison to read if the rebuild itself is the part you were hoping to skip.
There is a faster test than rebuilding a graph to find out whether you needed one. Describe the job to Gravity in one sentence and see whether an agent for it already exists. The first one is free, and there is no card and no server.
3. Dify: best if the flows were really RAG apps
Was your instance really document question-answering with a chat widget bolted to the front? Then Dify is the shorter rebuild, because retrieval is the product itself. Ingestion, chunking, reranking, prompt management and a publishable app are all first-class screens.
Read the licence before you build a product on it. Dify ships under its own terms, an Apache 2.0 base with added conditions covering multi-tenant resale and removal of branding. For internal use that changes nothing at all. If you intend to sell a hosted service built on it, that is a conversation with a lawyer.
What it does that Flowise never focused on
Evaluation and observability come in the box. You can see which chunks answered a question and why, which is the part every team eventually builds badly for itself.
Pros
- Retrieval, evaluation and prompt management are built in
- Free to self-host
- A free Sandbox tier on their cloud if you would rather not host
- Publishable apps and APIs at the end of the build
Cons
- Licence conditions on multi-tenant resale and branding
- Self-hosting is a heavier stack than a single Flowise container
- Weaker fit if your flows were general automation
- Best for
- The team whose Flowise instance was a knowledge base with a chat box in front of it.
- Pricing
- Self-hosting the community edition is free, plus your server and model tokens. Dify Cloud has a free Sandbox tier with paid tiers above it. I did not re-read those figures for this post, so price them on the vendor page before you commit.
- Free tier
- Yes, the Sandbox tier on Cloud, and self-hosting costs no licence fee.
- Verdict
- The shortest rebuild of the seven if your use case was retrieval. Gravity vs Dify sets it against the finished-agent model.
4. Activepieces: open source without owning the box
Activepieces answers the specific problem that brought you here, which is that owning a server is how you ended up with an unpatched CVE in the first place. The core is open source, and the same product runs on their cloud at the same price as it runs on your hardware. You pick which one, and you can change your mind.
The billing unit is the kindest on this page. One credit covers a whole run, whatever the step count, so you can price a flow before you build it. The free allowance refills daily, which keeps small flows alive without a card on file.
The shape is automation. Triggers, actions, branches, with AI steps available inside them. If your chatflows were agents with tools and memory, you will be re-expressing them, and that is a different mental model to learn on top of a migration you did not ask for.
Why I see this one up close
Gravity's builders publish their agents in Activepieces, so it is in front of me most working days. Take that as a bias. The credit model is the easiest here to budget: one number, one run.
Pros
- One credit per run, whatever it crosses
- Same price on their servers or yours, so hosting is a reversible choice
- The free allowance refills every day
- Open-source core with a commercial team maintaining it
Cons
- Smaller connector catalogue than Zapier or Make
- Plus stops at five seats, and the next tier is a long step
- Automation-shaped, so LLM-heavy flows need rethinking
- Best for
- The reader who wants to stay in open source but has decided, this week, that they are finished being the ops team.
- Pricing
- Free, 100 credits a day. Plus $16 a month billed yearly for 10,000 credits, then $0.007 a credit. Team $166 a month. Read on the Activepieces pricing page on 11 September 2026.
- Free tier
- Yes, and it refills every day.
- Verdict
- The best compromise between keeping the licence and dropping the pager. Gravity vs Activepieces explains where the two fit together.
5. LangGraph: the direction the sunset notice points at
Go back to the reason the Flowise team gave. They said the rigid low-code workflow approach hits a limit when things get complex, and that developers are moving to coding agents. Follow that argument to its destination and you arrive at a library.
LangGraph is an MIT-licensed Python library that treats an agent as a state machine: explicit nodes, explicit edges, explicit state, with pause-and-resume for human approval as a first-class feature. Our open-source AI agent frameworks comparison counted it at roughly 37,000 GitHub stars when we checked in July 2026.
You lose the canvas, and for a non-technical owner that single sentence ends the discussion. For a team already writing Python, losing the canvas is the point: version control, tests, code review and a diff you can read all come back, and the complexity ceiling the Flowise team described stops being a ceiling.
This is the trade the sunset notice was describing
Flowise's own team said the low-code shape runs out of room. Moving into code removes that limit and puts an engineer in the path of every future change. Price that cost now. Discovering it later is more expensive.
Pros
- MIT licence, no vendor between you and the code
- State, branching and human approval are native concepts
- Everything lives in version control, where it can be reviewed
- No canvas to maintain and no server unless you want one
Cons
- Nobody non-technical will touch it
- Observability, scheduling and retries become your build
- Heavier than the job needs if the flow was genuinely simple
Pick this if an engineer already owns the flows and they had outgrown the drag-and-drop shape before any of this happened. Read the frameworks comparison first, then write one flow in it before you commit the rest.
Worth reading alongside this: OpenAI Agent Builder alternatives covers the same migration question from the other end of the market, including why an SDK export does not convert a visual graph one for one.
6. Zapier Agents: nothing to host, ever
The opposite end of the range from a forked repository. There is no source code to own, no container to patch, and no repository that can be archived out from under you. You describe what the agent should do in plain English, connect accounts through an ordinary login, and it runs on somebody else's infrastructure.
The trade is complete and you should look at it directly. No source, no self-hosting, no portability, and a vendor who can change the product whenever they like. After the week you have just had, some readers will find that disqualifying and some will find it restful. Both reactions are rational.
What the catalogue buys you
What it buys is reach. Zapier's connector catalogue is the largest in this category, so the obscure app that had no Flowise node probably has a Zapier one. Agents sit inside the same account as your Zaps, and their allowance is listed separately from Zap tasks. I did not re-read the agent-specific limits for this post, so price your own volume on Zapier's page.
Pros
- Nothing to host, patch, fork or monitor
- The largest connector catalogue on this page
- A real free tier to test the idea
- Plain-English setup, so a non-technical owner can run it
Cons
- Closed source, no self-hosting, no export path
- Agent allowances are metered separately and need checking
- Long multi-step work gets expensive on per-task billing
It fits the owner-operator who has concluded that running infrastructure was never their job. The platform plans page shows a free tier with 100 tasks a month, Professional from $19.99 a month on annual billing, and Team from $69. Read on 11 September 2026.
This is the cleanest break with self-hosting available here, at the price of every right you had under Apache 2.0.
7. Gravity: for the reader who is done owning infrastructure
This is my product, so weigh it accordingly. It is last because it is the least like Flowise of anything on this page, and because I would rather you read the six above first.
Gravity is not open source and you cannot self-host it. If the Apache 2.0 licence was a real reason Flowise was in your stack, stop reading here and pick one of the six above. I would rather say that in the first two lines than have you discover it in week two.
What it does instead is remove the build. You type the task in plain words. Gravity matches it to an agent a builder has already written and tested, you sign in to the accounts it needs, you answer a few setup questions, and it runs on a schedule. The builder who wrote it keeps it working. That is their job, not yours, and that is the entire pitch after an archived repository.
Anything that messages a customer or moves money waits for your approval. Every agent passes automated checks before it is listed. Gmail, Sheets, Calendar, Drive, Slack, Notion and HubSpot connect through an ordinary login, so you never paste an API key into a config file you then have to protect.
The one argument for it after an archived repo
Flowise's wind-down handed you a maintenance job you never applied for. Gravity is the option where that job belongs permanently to someone else, and the price is that you never open the hood.
Pros
- Nothing to host, patch, fork or migrate
- Built and tested by someone who does this for a living
- Customer-facing steps pause for your approval
- One number a month, with a lot of usage included
Cons
- Not open source, and there is no self-hosting
- No blank canvas, only published tasks
- Live since September 2026, so the catalogue is still short
- Wrong answer if keeping your graph was the requirement
- Best for
- The solo operator or small team whose Flowise instance was doing one recurring job, and who never wanted a server in the first place.
- Pricing
- First agent free, no card. Then Autopilot at $20 a month worldwide, ₹1,999 a month in India, with a lot of usage included. India also has Minipilot at ₹399 a month with a smaller usage allowance. Buy more usage if you run out. As of September 2026.
- Free tier
- Yes. One agent, free.
- Verdict
- Right only for the reader who concluded from all this that they do not want to maintain infrastructure at all. Gravity vs Flowise puts the two models side by side.
Migration: what you actually carry over
Rebuild is the realistic path, whichever of the six you pick. Flowise chatflows export as JSON, and I know of no Flowise importer in any tool on this page. That JSON is a portable record of what you built, and nothing more than that.
The thinking transfers, and it is worth more than the file. Your prompts, your chunking choices, your tool list, your retry logic and the order of operations are all reusable in any of these. Copy them out of the JSON into a plain document before you open a new tool, because that document is the actual asset and it survives every future migration too.
Credentials do not transfer, and you will be re-entering every one. Custom nodes written against the Flowise runtime do not transfer either, because nothing else runs them. Anything that depended on a component unique to Flowise becomes a design decision.
Only a fork preserves your flows exactly. That is the strongest single argument for forking, and it is why the fork question belongs before the tool question.
One sequencing note to close on. Take the instance off the public internet first, rebuild second, and pick your winner at whatever pace the work deserves. The migration is not urgent. The network boundary is.
If the rebuild is the part you are dreading, check whether it is necessary at all. Run your first agent free on the task your chatflow was doing, and you will know inside a few minutes whether you needed a canvas or an outcome.
Which one should you choose?
Start from what you want to own after the move. Owning things is what just cost you a week.
Want the canvas and happy to keep a server running? n8n, with the fair-code caveat read properly, or Langflow if the LangChain component model was the appeal and MIT matters to you.
Documents and a chat box? Dify, and it is the shortest rebuild of the seven.
Open source, and you never want to patch a box again. Activepieces runs the same product on their cloud at the price of your own, and you can move between the two later.
An engineer already owns this and the flows had outgrown drag-and-drop before July. LangGraph is where you were heading anyway. The Flowise team's own reasoning agrees with you, which is a strange kind of endorsement but a real one.
If you never want to read another licence, Zapier Agents. You give up every right you had, and you get back the evenings.
If Flowise is load-bearing infrastructure you have already customised, and somebody who reads TypeScript will still be on the payroll in a year, fork it. Budget for the two open CVEs in the first sprint and treat the fork as a product you now maintain.
My own bias, stated plainly. For a recurring business task that a published agent already covers, I would pick the option where nobody builds or maintains anything. That is Gravity, the first agent is free, and if you want the cheaper end of the whole category instead, the cheapest AI agent platforms and the best platforms under $50 both include options I did not list here.
FAQ
Is Flowise shut down?
No. The GitHub repository was archived on 10 August 2026 and the core team's presence ended on 31 August 2026, but the project was not deleted and no service was switched off on a deadline. A self-hosted Flowise install keeps running exactly as it did; what ended is active maintenance.
Can I still use Flowise after 31 August 2026?
Yes. Nothing in the sunset notice expires your install, and the Apache 2.0 licence is not revoked by an archive. The cost of staying is that you are now running software with no upstream security releases, including two vulnerabilities published after the final version.
Is Flowise still open source?
Yes. The source remains on GitHub under Apache 2.0 and the sunset notice tells readers the code is theirs to keep building on. An archived repository is read-only for issues and pull requests, not withdrawn, so forking is both legal and expected.
What happened to the Flowise npm package and Docker images?
Both were marked deprecated on 10 August 2026, the same day the repository was archived. Existing versions can still be pulled, so an install or a rebuild will not fail today. Treat the deprecation warning as a signal that no fixed version is coming from that channel.
Is Flowise Cloud still running?
The sunset notice says nothing about the hosted product, and flowiseai.com was still serving a Pricing page when I read it on 13 September 2026. I will not claim the hosted service is dead, because the vendor has not said so. If you are a paying Cloud customer, ask them directly and get the answer in writing.
What is the closest alternative to Flowise?
Langflow is the nearest match to the visual builder shape, and it is MIT licensed and free to self-host. If your flows were really document question-answering with a chat front end, Dify is the shorter rebuild. Neither one can import a Flowise chatflow, and AI agent vs workflow automation explains why the shapes differ.
Should I fork Flowise or move to something else?
Fork if Flowise is infrastructure you have already customised and somebody on the payroll reads TypeScript, because a fork is the only path that preserves your flows exactly. Switch if Flowise was a convenience you could rebuild in a week. A fork inherits CVE-2026-67622 and CVE-2026-67620 on day one, and patching them becomes your job.
Sources
- Flowise, "Sunset notice", signed "Henry & Flowise Team". Read 13 September 2026. Source of the 29 July, 10 August and 31 August 2026 milestones, the Apache 2.0 forking statement and the quoted reasoning.
- GitHub, FlowiseAI/Flowise repository record. Read via the GitHub API on 13 September 2026:
archived: true, last push 13 August 2026, 55,456 stars. - GitHub, Flowise release history. Final release
flowise@3.1.4, published 29 July 2026. No later release listed. Read 13 September 2026. - CVE-2026-67622, insecure direct object reference in the OpenAI Assistants integration, CVSS 3.1 base 9.9, published 6 August 2026, affects versions through 3.1.4. Read 13 September 2026.
- CVE-2026-67620, server-side request forgery via omissions in
DEFAULT_DENY_LISTinhttpSecurity.ts, affects versions through 3.1.4. Read 13 September 2026. - Flowise pricing page, still serving as of 13 September 2026. Cited only to support the statement that the notice makes no claim about the hosted product.
- n8n pricing, read 11 September 2026.
- n8n, "Sustainable Use License" (LICENSE.md), the licence text in n8n's own repository. Source for the licence name, the fair-code characterisation and the internal-use rights described in the n8n section. Read 13 September 2026.
- Activepieces pricing, read 11 September 2026.
- Zapier plans and pricing, read 11 September 2026. Agent allowances are listed separately and were not re-read for this post.
- Langflow repository and Dify repository and LangGraph repository, for licence terms. Read 13 September 2026. Managed pricing for these three was not re-read for this post.
- Gravity pricing, September 2026.
