Free · No account
A compliance check inside your coding agent.
Building an app? Add Regulane to Claude Code, Cursor, Copilot, Codex, Windsurf or Gemini CLI with one line. Your agent checks your code against 34 secure-coding rules mapped to ISO 27001 and the DPDP Act, and you get a scorecard. Your code never leaves your machine.
Free · No account · Regulane never receives your code
- Secure-coding rules
- 34
- Areas, from secrets to infrastructure
- 8
- To add it to your agent
- One line
- Stored from a free check
- Nothing
01Set it up
One line in the agent you already use.
No token and no sign-up. Run the command, or add the lines to the file shown, then restart the agent if it asks.
Claude Code
Run in a terminal
claude mcp add --transport http regulane https://regulane.com/api/mcp
Cursor
~/.cursor/mcp.json, or .cursor/mcp.json in one project
{
"mcpServers": {
"regulane": {
"url": "https://regulane.com/api/mcp"
}
}
}VS Code (Copilot)
Run in a terminal
code --add-mcp '{"name":"regulane","type":"http","url":"https://regulane.com/api/mcp"}'Codex
Run in a terminal
codex mcp add regulane --url https://regulane.com/api/mcp
Windsurf
~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"regulane": {
"serverUrl": "https://regulane.com/api/mcp"
}
}
}Gemini CLI
Run in a terminal
gemini mcp add --transport http regulane https://regulane.com/api/mcp
Other agents
Any agent that runs only local (stdio) servers, through the mcp-remote bridge
{
"mcpServers": {
"regulane": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://regulane.com/api/mcp"
]
}
}
}The address is https://regulane.com/api/mcp, a Model Context Protocol server over HTTP.
02How a check runs
Your agent does the reading. Regulane supplies the rules.
Regulane never runs or receives your code. It tells your agent what to look for, and turns what the agent found into a scorecard.
- 01Ask your agentYou“Check this app with Regulane.” Clients that support MCP prompts also get a compliance_review command.
- 02Sends the rulesRegulane34 secure-coding rules, each saying what to look for and how to fix it, mapped to ISO 27001 Annex A and the DPDP Act.
- 03Reads your code, on your machineYour agentThe agent you already use checks the code it already has open. Regulane never receives the code.
- 04Reports what it foundYour agentFor each problem: a file path, a line number and one sentence. Anything that looks like a secret or personal data is removed on arrival.
- 05Returns a scorecardRegulaneFindings by severity, the controls they affect, the fix for each, and what was not checked. A free check is not stored.
03The scorecard
What comes back to your agent.
This is the real output for an example app with three findings. Your agent shows it to you and can offer to make each fix.
Regulane compliance check of github.com/example/expense-app (main), ruleset 2026.10
Rules checked: 29 of 34.
Result: 3 findings: 1 critical, 2 high.
Affects ISO 27001 controls A.5.15, A.5.34, A.8.3, A.8.11, A.8.15, A.8.24, A.8.26 and DPDP Act s.8(5).
Findings, most severe first:
- [critical] RG-SC-004 Record fetched without an ownership check — src/api/expenses/[id].ts:18
An expense is loaded by id without checking it belongs to the signed-in user.
Fix: Scope every query by the caller's tenant or owner from the session, for example where: { id, tenantId }, and answer not found for someone else's record.
- [high] RG-SC-018 Personal data in logs, errors or analytics — src/lib/auth.ts:52
Failed sign-ins log the full email address.
Fix: Log ids, not people. Mask or drop personal fields before logging, and scrub them in the error tracker's before-send hook.
- [high] RG-SC-016 Weak or home-made cryptography — src/lib/reset.ts:7
Password reset codes come from Math.random.
Fix: Use the platform's crypto library: AES-256-GCM or ChaCha20-Poly1305 with random nonces, SHA-256 or better, HMAC for integrity, and a CSPRNG (crypto.randomBytes, secrets) for anything secret.
Not checked: Dependencies and build pipeline; Infrastructure and configuration. Check these too for a complete result.
Regulane did not store this report.
To keep each repository's history, track and accept findings, and show them to an auditor as evidence for ISO 27001 and the DPDP Act, create a workspace at https://regulane.com/get-started and connect with an agent token.04The rules
34 rules, each tied to a control.
Ruleset 2026.10. 7 are critical. Every rule names the ISO 27001:2022 Annex A controls it relates to, and 7 also name sections of the DPDP Act 2023. Your agent receives each rule’s full check and fix.
Secrets, authentication and access control
- Secret committed to sourceRG-SC-001 · A.5.17, A.8.24, A.8.28critical
- Server secret exposed to the browser or app bundleRG-SC-002 · A.8.12, A.8.28critical
- Endpoint without authenticationRG-SC-003 · A.5.15, A.8.5, A.8.26critical
- Record fetched without an ownership checkRG-SC-004 · A.5.15, A.8.3, A.8.26critical
- Identity, role or outcome taken from the clientRG-SC-005 · A.8.26, A.8.28high
- Weak password storageRG-SC-006 · A.5.17, A.8.5, A.8.24critical
- Unsafe session or token handlingRG-SC-007 · A.8.5, A.8.24high
- No limit on sign-in or costly endpointsRG-SC-008 · A.8.5, A.8.6medium
- Security control switched offRG-SC-034 · A.8.28, A.8.32high
Injection and unsafe input handling
- SQL or NoSQL injectionRG-SC-009 · A.8.28critical
- Command injection or unsafe code executionRG-SC-010 · A.8.28critical
- Cross-site scriptingRG-SC-011 · A.8.28high
- Unsafe file path or upload handlingRG-SC-012 · A.8.28high
- Server-side request forgeryRG-SC-013 · A.8.20, A.8.28high
- Open redirectRG-SC-014 · A.8.28medium
- State change open to cross-site requestsRG-SC-015 · A.8.26, A.8.28high
Cryptography and transport security
- Weak or home-made cryptographyRG-SC-016 · A.8.24high
- Transport security disabledRG-SC-017 · A.8.20, A.8.24high
Personal data (DPDP Act)
- Personal data in logs, errors or analyticsRG-SC-018 · A.5.34, A.8.11, A.8.15 · DPDP s.8(5)high
- Personal data collected without notice or consentRG-SC-019 · A.5.34 · DPDP s.4, s.5, s.6medium
- Personal data kept with no way to erase itRG-SC-020 · A.5.34, A.8.10 · DPDP s.8(7), s.12medium
- Children's data processed unsafelyRG-SC-021 · A.5.34 · DPDP s.9high
- Sensitive identifiers stored or shown unprotectedRG-SC-022 · A.5.34, A.8.11, A.8.24 · DPDP s.8(5)high
- Personal data sent to a new third partyRG-SC-023 · A.5.19, A.5.20, A.5.23 · DPDP s.8(2), s.16medium
- Personal or confidential data sent to an AI modelRG-SC-024 · A.5.34, A.8.11, A.8.12 · DPDP s.8(5)medium
Logging and error handling
- Security event not recordedRG-SC-025 · A.8.15, A.8.16medium
- Error response leaks internalsRG-SC-026 · A.8.12, A.8.28medium
Dependencies and build pipeline
- Risky dependency addedRG-SC-027 · A.5.21, A.8.8, A.8.19high
- Build pipeline can be tampered withRG-SC-028 · A.5.21, A.8.25, A.8.32medium
Infrastructure and configuration
- Insecure infrastructure configurationRG-SC-029 · A.5.23, A.8.2, A.8.9, A.8.20high
- Permissive CORS or security headers removedRG-SC-030 · A.8.9, A.8.26medium
- Production data or credentials outside productionRG-SC-031 · A.8.11, A.8.31, A.8.33medium
Change safety and testing
- Destructive data change without a safeguardRG-SC-032 · A.5.33, A.8.13, A.8.32medium
- Security change without a test of the deny pathRG-SC-033 · A.8.29low
05Your code stays with you
What Regulane receives, and what it never does.
What your agent sends
- The repository name or address (any credentials in it are removed), branch and commit
- Which agent ran the check, and one line on what it looked at
- The rule ids it checked
- For each finding: a file path, a line number and one sentence
What Regulane never receives or keeps
- Your source code, diffs or file contents
- Secrets: anything that looks like a key, token or password is replaced before the report is read
- Personal data: email addresses, card, Aadhaar and PAN numbers are replaced the same way
- Any record of a free check: the scorecard goes back to your agent and is not stored
What a free check is not
- Regulane does not run your code or scan it itself. Your coding agent does the checking, so the result is only as careful as the agent and the scope you give it.
- A clean scorecard is not a certification, an audit opinion or a penetration test.
- The rules cover secure coding and personal-data handling in code. They do not cover your policies, people, suppliers or physical security, which an ISO 27001 audit also looks at.
- A free check keeps no history. To track findings over time, accept them with a reason and show them to an auditor, connect a Regulane workspace.
06For a team
Keep every check as evidence.
Connect a Regulane workspace with a token per developer, and each repository’s checks and findings are kept for your team and your auditor.
14-day trial · No card required
| What you get | Free check | Workspace |
|---|---|---|
| All 34 rules | Yes | Yes |
| Scorecard in your agent | Yes | Yes |
| Account needed | No | Yes |
| History of every check, per repository | No | Yes |
| Open findings your agent can come back to | No | Yes |
| Accept a finding with a reason, in the audit trail | No | Yes |
| A token per developer, revoked when they leave | No | Yes |
07Make it automatic
Checked on every change, without asking.
Add this to your repository’s AGENTS.md (read by Codex, Cursor, Copilot, Windsurf and Gemini CLI) or CLAUDE.md (Claude Code).
AGENTS.md
At the root of your repository
## Compliance checks This repository is checked with Regulane. Before finishing a change that touches authentication, access control, personal data, cryptography, logging, dependencies, CI or infrastructure, use the `regulane` MCP server: call get_compliance_rules, check the change against those rules, and record the result with report_compliance_check. Never include source code, secrets or personal data in a report.
Questions
About the free check
Is the check really free?
Yes. Connect without a token and the rules and the scorecard cost nothing and need no account. Keeping the history of checks in a Regulane workspace is part of the Intelligence plan, and a trial includes it.
Does Regulane see my code?
No. Your coding agent reads the code on your machine, under your agreement with the agent’s provider. Regulane sends it the rules and receives only what it reports: file paths, line numbers and one-sentence findings. Anything in those that looks like a secret or personal data is replaced on arrival.
Which coding agents does it work with?
Any agent that supports the Model Context Protocol over HTTP: Claude Code, Cursor, VS Code with GitHub Copilot, Codex, Windsurf and Gemini CLI, each with the setup shown on this page. Agents that only run local servers can use the mcp-remote bridge.
What do the rules cover?
34 rules in 8 areas: secrets and access control, injection, cryptography, personal data, logging, dependencies and build pipelines, infrastructure configuration, and change safety. Each names the ISO 27001:2022 Annex A controls it relates to, and 7 also name sections of the DPDP Act 2023.
Is a clean scorecard a certification?
No. It says your agent checked the code it was given against these rules and found nothing. It is not an audit opinion or a penetration test, and Regulane does not run or scan your code itself.
Can my team keep the results as audit evidence?
Yes, in a Regulane workspace. A company administrator or manager creates an agent token per developer, and every check is kept per repository with its findings, which managers can accept with a reason. They are shown as agent-reported and do not count toward ISO readiness.
Keep reading
- ISO 27001Annex A, clauses 4–10, the Statement of Applicability and audit preparation.
- DPDP ActThe Data Fiduciary obligations, the processing register, requests and breaches.
- SecurityWhere your data lives, what the code enforces, and what we do not claim.
- Regulatory IntelligenceWhich regulatory changes name something you run, and why.
Regulation changes every week. Your compliance programme should know what changed, and what to do next.
14-day trial · No card required · Per workspace, never per seat
