MintHCM: an AGPL-licensed HR platform built on the SuiteCRM codebase, with MCP and A2A interfaces for agents
Open-source HCM system for managing HR processes with AI agents. Full data ownership, MCP/A2A-native, no vendor lock-in.
At a glance
- What is it?
- MintHCM is a PHP human capital management system derived from SugarCRM Community Edition and SuiteCRM, shipped under AGPL-3.0 with MCP, WebMCP and A2A agent interfaces. The interesting question is not whether it has HR features, but whether a CRM codebase is the right foundation for HR data and what the AGPL means for the integrators it targets.
- Who is it for?
- MintHCM fits organisations that already run PHP and MySQL, need the HR data to sit on their own infrastructure, and have developers who will treat the source tree as the product. It does not fit teams that want a hosted HR suite, a published security response process, or a vendor contract with named support obligations.
- Can I use it commercially?
- Yes, with strict conditions. AGPL-3.0 is a network copyleft licence: if people use a modified version over a network, for example as a hosted service, you must offer them its source code under the same licence.
- Is it still maintained?
- Yes. The repository last received commits 27 days ago.
- What is it written in?
- Mainly PHP, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
The gap MintHCM claims: HR data you host, with agents as first-class clients
The README frames the market in three buckets. Hosted HR SaaS gives up data sovereignty and deep customisation. Other open source HR tools give control but, in its words, no native AI agent ecosystem. AI-first SaaS gives strong AI features without ownership of data or infrastructure. MintHCM positions itself in the fourth slot: open source plus agent interfaces.
That framing is a marketing construction, and you should read it as one. But the underlying claim is checkable. The repository is AGPL-3.0, the README states there are no gated features and no phone-home telemetry, and the deployment section lists on-premise, public cloud, private cloud and hybrid as supported targets. For an organisation in a jurisdiction that restricts where employee records may be stored, or one that simply refuses to have a payroll-adjacent system outside its own network, those are the properties that matter. The audience named in the README is enterprises with complex HR processes, organisations adopting AI in operations, companies needing full control over data and infrastructure, and system integrators.
The last group is the honest one. This is not a product you install and hand to an HR administrator on a Monday. It is a platform you build on.
What the codebase actually is: SugarCRM CE and SuiteCRM underneath
The README is unusually direct about lineage. MintHCM is based on SugarCRM Community Edition (referred to there as SugarAI) and SuiteCRM, and the project's own summary line is that MintHCM is CRM reinvented. The API section says the same thing from the other direction: MintHCM is based on SuiteCRM, so the API is very similar, with a few important differences covered in a separate guide.
This has practical consequences that the README does not spell out. SugarCRM and SuiteCRM use a metadata-driven data model: module definitions, field definitions, layouts and relationships live in declaration files rather than in hand-written SQL migrations. Customisations are typically packaged as modules that can be installed and uninstalled. Anyone who has worked with SuiteCRM Studio or its module loader will recognise the shape of MintHCM's extensibility claims about a flexible data model and modular architecture. Anyone who has not will find the learning curve is really the SuiteCRM learning curve.
The HR content on top is broad. The README lists recruitment and candidate management, onboarding and offboarding, employee profiles and employment history, competences and evaluations, time and attendance, leave management, performance and development, organisational structure, job descriptions and employer branding, resource booking and workplace management, and reporting. It also points to iOS and Android apps. What the README does not provide is a data model reference, an entity listing, or a description of how, say, a leave record relates to an employee profile and an attendance entry. For a system whose selling point is data ownership, that documentation gap is the first thing an evaluating team will hit.
MCP, WebMCP and A2A: what the README specifies and what it does not
The agent story is the differentiator, so it deserves the closest reading. The README states that MintHCM is designed to work within modern AI ecosystems and lists four interface families: MCP and WebMCP for connecting HR data and processes to MCP-compatible agents; A2A for orchestrating multi-agent workflows across systems; A2UI and AGUi as human-to-agent collaboration interfaces; and a built-in AI agent that is ready to use out of the box, plus support for custom agents built on HR data and processes.
It also states that the system is LLM-agnostic, naming OpenAI, Anthropic, Mistral, local Llama and regional sovereign models as options, with no forced dependency on a single AI vendor. That is a meaningful design commitment if it holds, because it means the agent layer is a client of the HR data rather than a wrapper around one model provider.
What is missing is the part an engineer needs. There is no MCP server endpoint path, no transport description, no tool or resource naming convention, no example call, and no A2A agent card example. The README asserts the architecture; it does not demonstrate it. The repository is the place to verify this, and the topics list includes mcp and mcp-server, which suggests there is something concrete in the tree. Treat the README's agent section as a statement of intent until you have read the corresponding code. The claim that agents can automate HR decisions rather than just workflows is the strongest claim in the document and the one with the least supporting detail.
Getting it running: PHP 8.2, MySQL 8.0, Apache, and a guide off-repo
The requirements section is short and specific. A Linux or Windows machine running Apache2. PHP 8.2. MySQL 8.0. That is the whole list in the README, and it is worth noting what it does not mention: no minimum memory, no required PHP extensions, no cron or queue component, no Redis, no search engine. SuiteCRM-derived systems typically need more than that list implies, so the omission is a documentation gap rather than evidence of a small footprint.
The README does not give an installation command sequence. It points to a guide at minthcm.org/support/minthcm-installation-guide/ and states that the installation process is described there. There is a Docker image referenced in the badge block at hub.docker.com/r/minthcm/minthcm, and the README's badge markup uses the tag minthcm/minthcm, but the README body never documents a docker run invocation, a compose file, or the environment variables the image expects. If you intend to evaluate MintHCM, the container image is the fastest path and the least documented one.
For the API, the README says it is very similar to SuiteCRM's with important differences, and links to minthcm.org/support/how-to-use-mint-api/. There is also an open API mentioned under extensibility, without a specification. Translation packages are managed through a Crowdin project, and there is a Discord server and a forum for support. None of these links come with a version guarantee, so the documentation you read may not match the 4.3.2.2 tree.
The AGPL-3.0 boundary and the upgrade cadence
The licence is AGPL-3.0, stated in the repository metadata and in the README's open source section. The AGPL's defining feature is the network clause: if you modify the software and let users interact with it over a network, you are generally required to offer those users the corresponding source. That is precisely the scenario MintHCM targets, since it is a server application accessed by employees. This is not legal advice and the obligations depend on your specific deployment and modifications, but the practical point is that AGPL is a deliberate choice here, not an accident, and it is consistent with the README's claim that there is no enterprise-only fork. If your organisation cannot accept copyleft obligations on a modified internal system, this is a reason to stop before you start rather than after.
The release history shows a steady cadence: 4.3.2 in July 2026, 4.3.2.1 later that month, and 4.3.2.2 in August 2026. The four-part version numbering is inherited from the SuiteCRM lineage. Because MintHCM is a derivative of two upstream projects, upgrades carry two kinds of cost: MintHCM's own release changes, and the drift between MintHCM's copy of the SugarCRM CE and SuiteCRM code and whatever those upstreams have done since. The README does not describe a rebase or merge policy, so a team that customises the data model heavily is accumulating work whose size is not documented anywhere in the material. The README's own appeal for help, in the paragraph about how much goes into developing this class of business software, is a signal about the project's resourcing.
Where MintHCM is the wrong tool
The clearest failure mode is organisational rather than technical. MintHCM expects you to run Apache, PHP 8.2 and MySQL 8.0 yourself, and to treat the source tree as the system of record for your customisations. A company without PHP developers on staff, or one whose HR team needs a vendor support contract with response times, will spend more on the platform than it saves. The README's own target list includes system integrators and development teams for a reason.
The second is scope. MintHCM covers HR processes. It is not a payroll engine, and the README's capability list does not claim to be one. If your evaluation criteria include statutory payroll calculation, tax filing, or jurisdiction-specific benefits administration, this system does not address them and you would be pairing it with something else.
The third is documentation maturity relative to the ambition. The README describes MCP, WebMCP, A2A, A2UI and AGUi in a single bullet list, with no interface references. It describes an open API without a specification. It gives a requirements list of three items for a SuiteCRM-derived PHP application. For a project whose pitch is that you own and control everything, the amount of verification left to the reader is high. That is not disqualifying, but it changes the shape of the evaluation from a trial install into a code review.
The alternative to weigh it against: a hosted HR suite
The obvious comparison is a commercial SaaS HCM such as BambooHR, Workday or Personio. The difference is not feature count, it is where the data lives and who holds the upgrade lever. In a hosted suite, the vendor runs the database, ships the schema changes, and absorbs the operational work. You get a support contract, a compliance programme, and a documented API with published versioning. You give up the ability to change the data model, to run the system in a jurisdiction of your choosing, and to keep working if the vendor changes its pricing.
MintHCM inverts every one of those. You run MySQL 8.0 and PHP 8.2 on your own Apache host. You change the data model in code. You decide where it is deployed. In exchange, you own the patching, you own the upgrade path across a fork of a fork, and you own the security response when something in the SuiteCRM lineage needs fixing. The README's claim of no phone-home telemetry is the concrete expression of that trade: nothing leaves your infrastructure, including the telemetry that would otherwise help the project understand its own usage.
The second alternative is plain SuiteCRM itself, without the HR layer. If your need is a CRM with some HR-adjacent records, adopting MintHCM means taking on an additional derivative layer for capabilities you may not use. If your need is genuinely HR, the reverse applies: MintHCM is the layer that turns the CRM modules into recruitment pipelines, onboarding checklists and leave records.
Editorial conclusion
MintHCM fits organisations that already run PHP and MySQL, need the HR data to sit on their own infrastructure, and have developers who will treat the source tree as the product. It does not fit teams that want a hosted HR suite, a published security response process, or a vendor contract with named support obligations. Before committing, read the installation guide at minthcm.org/support/minthcm-installation-guide/, confirm the exact PHP 8.2 and MySQL 8.0 versions you will run, and check whether the MCP and A2A interfaces you need are actually implemented in the 4.3.2.2 source tree rather than only described in the README.
Community notes