EasySpecs
EasySpecs maps a Git repository into functional documentation, then turns raw change requests into readable specifications paired with Trust Specs — validators and evals defined before coding agents generate any code.
What is EasySpecs?
EasySpecs describes itself as a Spec Engineering Assistant and a Documentation Factory as a Service. Its premise is that when coding agents generate faster than humans can review, quality has to be fought for in the specification rather than in code review. The product connects to a Git repository through a read-only personal access token on GitHub, GitLab, Azure DevOps or Bitbucket, and works in three stages.
The first stage produces functional documentation of the system as it actually behaves, claiming up to 98% line-of-code coverage assignment, and exposes it as a knowledge tree spanning features, use cases, scenarios, views, interactions, services, methods, data models, fields, relationships, tech stacks, infrastructure, components, tests and test cases. The second stage takes a raw, human-written change request and expands it into intent grounded in that documentation, then into a diagram, then into a detailed specification with enough substance for an agent to plan and code against. Sub-specifications open directly from any diagram node.
The third stage is what the vendor calls Trust by Design. Every specification is paired with a Trust Spec carrying validators, evals and checks: a hard in-and-out scope boundary, and numbered behavioural claims each with a False if condition. Unresolved decisions stay marked not executed or blocked rather than passing by default.
Two surfaces are offered — a browser workspace for stakeholders and an IDE extension for Visual Studio Code, Cursor and Antigravity, which pulls pre-processed context and change requests, and pushes edited specifications back. The tool addresses developers, technical product managers and organisation leaders, and positions itself upstream of coding agents such as Codex, Claude Code and Cursor rather than as a replacement for them. The site also hosts a large editorial hub on agentic coding, plus separate consulting and training lines that are not part of the product.
What it does
- Map a connected Git repository into functional documentation, with up to 98% line-of-code coverage assignment
- Browse a knowledge tree covering features, views, services, data models, tech stack, infrastructure and tests
- Expand a raw change request into grounded intent, then a diagram, then a specification an agent can act on
- Attach a Trust Spec to every specification: scope boundary and numbered claims with a False if condition
- Track every change request and its linked specification across the whole organisation
- Design agentic flows visually, combining programmatic and agent steps with programmatic and AI-judge evals
- Work from Visual Studio Code, Cursor or Antigravity: pull context and docs, edit specs, push changes to the cloud
When to use EasySpecs / When not to
A quick filter to help you decide if EasySpecs is the right fit.
When to use EasySpecs
- Development teams already practising Spec-Driven Development alongside coding agents
- Technical product owners who write specifications engineers should not have to reinterpret
- Engineers inheriting an undocumented or legacy codebase nobody on the team wrote
- Consultancies, agencies and freelancers who must scope a client repository before committing to dates
- Teams weighing a migration or rewrite, or running technical due diligence on a codebase
When not to use EasySpecs
- Anyone expecting the tool to write the code itself — it produces specifications that other agents consume
- Buyers with hard compliance requirements: no privacy policy, no DPA and no certification are published
- Teams needing a mobile app, as none exists on either the App Store or Google Play
- Developers wanting to drive the platform programmatically, since no public API is documented
- Organisations that need a permanent free tier, absent from the current three-plan price list
How to use EasySpecs
A typical end-to-end flow, from setup to results.
- Sign in to open your projects and the browser workspace
- Install the EasySpecs extension in Visual Studio Code, Cursor or a compatible editor, from the editor's extension marketplace or via the published one-line curl installer
- Create a read-only personal access token on GitHub, GitLab, Azure DevOps or Bitbucket and enter it in Settings
- On the Workbench plan, add your own inference keys in Settings (OpenRouter, OpenAI, Anthropic, Google or another supported provider)
- Point the extension at your repository and run the analysis to build a first snapshot
- Review the functional map in the browser and explore the knowledge tree of your system
- Enter a change request, then let it be expanded into grounded intent and a diagram
- Write the specification in structured and HTML views, splitting it into sub-specs where useful
- Add Spec of Trust validators, the scope boundary and the numbered behavioural claims
- Send the context and specification to your coding agent, or start coding from the IDE extension
Pros & Cons
Pros
- Read-only repository access, explicitly bounded: the vendor states it does not push commits or change your repo
- Low published entry point at 10 EUR per repository per month, with a full public comparison of the three plans
- Bring-your-own-key inference on Workbench, with an explicit commitment not to add a markup on provider rates
- A precise, published coverage figure rather than a vague claim: up to 98% line-of-code coverage assignment
- Credits on the Factory plan are organisation-wide and can be spent on whichever repository needs them
- A genuinely published, versioned IDE extension under an MIT licence, with a visible download count
- Self-hosting is offered on the Industrial plan, on VPC, private cloud or on-premise infrastructure
Cons
- No privacy policy is published at all, which is a serious gap for a product that reads customer source code
- The terms are visibly stale: dated 25 March 2025, they describe the product as a free tool while three paid plans are sold
- The terms name Spaii Edutainment SL while the footer of every page names Spaii Labs SL — two distinct legal entities
- No data processing agreement, no subprocessor list, no retention period and no certification are published
- No public API is documented, so the platform cannot be driven programmatically
- The backend at api.easyspecs.ai was returning errors during review, leaving the credit conversion table unreadable
- The advertised entry price is a limited-time 50% promotion with no published end date; the reference rate is double
Pricing & Plans
There is no permanent free plan and no free trial on the current price list. The cheapest paid entry point is the Workbench plan at 10 EUR per repository per month, currently advertised at 5 EUR under a 50% limited-time promotion with no published end date. Workbench operates on a bring-your-own-key basis, so model inference is billed separately by the customer's own provider; the vendor states it adds no markup on provider rates. The Factory plan costs 150 EUR per month and includes fifteen organisation-wide credits usable on any connected repository, with additional blocks of ten credits available at 150 EUR. The Industrial plan is quoted individually. It should be noted that the terms of 25 March 2025 still describe the product as free, contradicting the pricing page.
- one connected Git repository
- authoring in browser and IDE
- and bring-your-own-key inference billed by your own provider. No credits are used.
- everything in Workbench
- plus fifteen organisation-wide credits each month spendable on any connected repository
- with inference covered by vendor-managed keys. Extra credits cost 150 EUR per block of ten.
- deployment on your own infrastructure (VPC
- private cloud or on-premise)
- bespoke token
- credit and procurement terms
- aligned with security and data-residency requirements.
Data, GDPR & hosting
A consolidated view of how EasySpecs handles your data.
GDPR overview
GDPR implementation is partial and thinly documented. The vendor is established in Barcelona, Spain, applies Spanish law and names Barcelona courts as competent. The terms cite the GDPR and EU ePrivacy rules when describing consent for non-essential cookies, and name standard contractual clauses and adequacy decisions as transfer mechanisms for processing outside the European Economic Area. Beyond that, the gaps are substantial and worth stating plainly: no privacy policy is published anywhere on the site, no data protection officer or Article 27 representative is designated, no dedicated privacy contact address is given, no subprocessor list is published despite subprocessors being acknowledged, no data processing agreement is offered, no retention period appears, and no data subject right — access, erasure, portability, objection — is set out. Compliance is never claimed in so many words.
Who owns the data?
The terms state that customers keep the rights to their own content, granting the vendor only the rights reasonably necessary to operate and improve the product — hosting, processing and displaying what is submitted. The product itself, its branding and the vendor's own content belong to Spaii Edutainment SL or its licensors. Repository access is deliberately narrow: connection uses a read-only personal access token, and the pricing page states the token serves to read the codebase for documentation production, not to push commits or alter the repository. No separate privacy policy exists, so these terms are the only place where data rights are addressed at all.
Reuse rights
Customers keep their own content and are not asked to seek permission to reuse it; the licence they grant runs the other way, and is limited to operating and improving the service. On the vendor's side, the terms describe cookies, local and session storage, analytics, diagnostics and third-party tracking scripts, pixels or SDKs used to measure traffic, conversions, product quality and marketing effectiveness. Identifiers named include IP address, device and browser type, approximate location derived from IP, referral URLs and on-site behaviour. Data may be processed by subprocessors or partners outside the European Economic Area under standard contractual clauses or adequacy decisions. On the Workbench plan, model inference runs on the customer's own provider keys.
Data retention & training
Hosting summary
The site makes no hosting declaration: no country and no region is named as the place where customer data is stored. The only related statement sits in the terms, which say data may be processed by subprocessors or partners outside the European Economic Area, relying on standard contractual clauses or adequacy decisions. That is a transfer clause, not a hosting commitment, and it should not be read as one. Technical observation during review, independent of any vendor statement, showed the marketing site served through Amazon CloudFront and the application backend resolving to an Amazon EC2 instance in the eu-west-1 region in Ireland; neither is a contractual guarantee. Customers with data-residency requirements are directed to the Industrial plan, which offers deployment on the customer's own infrastructure — VPC, private cloud or on-premise — described as aligned with security and data-residency requirements. Anyone with a jurisdiction constraint should obtain the hosting location in writing.
Things to keep in mind
Risks and trade-offs to weigh before adopting EasySpecs.
- The publisher's identity is contradictory: the terms name Spaii Edutainment SL while every page footer credits Spaii Labs SL. Both are active Spanish companies sharing a registered office and a sole director, but only the second has software consultancy as its registered activity. Check which entity you are actually contracting with before signing.
- The terms are stale and therefore unreliable as a contract: dated 25 March 2025, a year before the domain itself was registered, they still describe the product as a free tool while three paid plans are on sale. Any commercial commitment you infer from them is revocable.
- No privacy policy exists anywhere on the site, and no subprocessor list, retention period or data processing agreement is published. For a tool that ingests your source code, that is a governance gap you should resolve in writing before connecting a client repository.
- The advertised 5 EUR price is a 50% limited-time promotion with no published end date; the reference rate is 10 EUR per repository per month. Budget on the reference rate, and remember that on Workbench you also pay your own inference bill separately.
- Treat generated documentation and specifications as drafts to be reviewed, not as verified truth. A knowledge tree covering 98% of lines is an impressive coverage figure, not a correctness guarantee, and a confident-looking map of an unfamiliar codebase is exactly the kind of artefact people stop questioning.
- The falsifiable claims and validators are only as good as the person who reviews them. There is a real risk of transferring judgement to the tool — of approving a Trust Spec because it looks rigorous rather than because you checked the claims against what the system is supposed to do.
- Service continuity is not proven: the backend was returning errors during review, leaving part of the pricing page unreadable, and a development artefact pointing at localhost was still live on the installation page. Both suggest a product still stabilising.
Setup & Integrations
Technical difficulty
Moderate, and clearly aimed at technical users. Setup is documented in six steps: create an account, install the IDE extension, connect a repository with a read-only personal access token you generate yourself, and run the analysis. The extension installs either from the editor's marketplace or through a one-line curl installer that requires curl and a supported editor CLI on your PATH. Workbench users must additionally create and enter their own inference provider keys. No server work is needed on Workbench or Factory; the Industrial plan involves deploying on your own infrastructure.
Deployment
Integrations
Behind EasySpecs
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement EasySpecs.
Frequently asked questions
What does EasySpecs actually do?
How do I connect my code, and what access does the tool get?
How much of my codebase gets documented?
What does it cost, and is there a free plan?
Do I have to supply my own AI keys?
Is there an IDE extension, and where do I get it?
Is there a mobile app or a public API?
Can I run it on my own infrastructure?
What are the GDPR and data protection guarantees?
Who is behind the tool?
Should you pick EasySpecs?
EasySpecs is a young but genuinely shipped product with an unusually clear thesis: if agents write the code, the place to enforce quality is the specification, not the pull request. The three-stage path from repository to functional documentation, from documentation to grounded intent, and from intent to a specification paired with falsifiable claims is coherent and well documented. The Trust Spec idea — deciding how you will know a change works before an agent writes it — is the part that distinguishes it from generic documentation generators. The published pieces back this up: a detailed public price list, eleven documented use cases, and a versioned IDE extension available under an MIT licence.
The weaknesses sit almost entirely on the trust and governance side, which is awkward for a tool whose whole pitch is trust. No privacy policy is published at all, for a product that reads customer source code. The terms are a year old, describe the product as free while three paid plans are sold, and name a different legal entity from the one credited in the site footer. There is no data processing agreement, no subprocessor list, no retention period and no certification. During review the backend was returning errors, which left the credit conversion table unreadable — so the real cost of an action on the Factory plan could not be checked. A development artefact referencing localhost was also still live on the installation page.
It is worth a look for teams already committed to spec-driven development and agentic coding, where the entry price is low enough to test on a single repository. Teams buying under compliance constraints should wait for the legal documentation to catch up with the product.
- Choosing a selection results in a full page refresh.
- Opens in a new window.