Review a pull request
Supply the changed code and surrounding contracts. Focus the review on data crossing trust boundaries, permission checks and sensitive operations.
AI code security / review to remediation
Review code with security in mind.
Bring a diff, a request handler or a focused module to the LLMsRelay API. Ask it to follow data flow, check authorization assumptions and propose fixes that your developers can review and test.
Sign in or create an account. Continue directly to the Uncensored wallet.
Put AI to work on the task
From a focused diff to a reviewable fix.
Understand a potential weakness
Request the relevant function, line reference and triggering conditions for each concern. Separate confirmed code behavior from missing runtime context.
Plan remediation
Ask for a minimal fix and tests that would detect the regression. Review compatibility, behavior changes and assumptions before applying the patch.
A workflow you can verify
- 1
Select the relevant code
Include caller context, input validation, roles and dependencies where they matter. Remove embedded credentials before sending source.
- 2
Ask for traceable findings
Require file or excerpt references, data-flow reasoning and confidence limits. Treat the output as a review draft alongside your normal checks.
- 3
Verify the patch
Have a developer review the suggested fix, run the relevant tests and verify the behavior. Feed back failures with the actual output.
Example prompt
Review this code diff and its caller context for security issues. For each concern cite the relevant function or line, explain the trust boundary and preconditions, and separate confirmed behavior from assumptions. Suggest a minimal fix and regression test.
Ask for a useful result
- Findings tied to code references
- Assumptions and missing context
- A fix proposal with regression test ideas
Add another review pass to the code you ship.
Connect your script or IDE client to the API and request a focused review before your existing testing and approval process.
One wallet. Usage you control.
One wallet. Usage you control.
Choose a prepaid pack in your dashboard. The $500 and $1,000 packs include extra usage balance.
You pay
$45
Usage balance
$45
You pay
$100
Usage balance
$100
You pay+20% Bonus
$500
Usage balance
$600
You pay+30% Bonus
$1000
Usage balance
$1300
| Model | Context | Input | Output |
|---|---|---|---|
| glm-5-3-uncensored | 262,144 | $4 | $12 |
| glm-5-3-uncensored-1m | 1,000,000 | $12 | $20 |
Rates depend on the selected model. Input, output and cached tokens are billed separately; a dollar balance is not a fixed number of tokens.
Connect your own workflow
- 1Top up the Uncensored wallet in your existing LLMsRelay account.
- 2Create an Uncensored API key in the dashboard.
- 3Set https://api.llmsrelay.com/v1 in your OpenAI-compatible client and choose a public model ID.
Before you connect
Does this replace SAST or a security audit?
No. A model review is an additional analysis step. It can miss vulnerabilities or report false positives. Keep static analysis, tests and human review in your process.
Can it review an entire repository automatically?
The API processes the content your client sends. It does not automatically clone a repository or inspect omitted files. Select the relevant context, or implement retrieval in your own client.
Which model should I select?
Use glm-5-3-uncensored for the standard route or glm-5-3-uncensored-1m for larger text context. The 1M route is text-only and has a different token price. Check availability with your account key.
Explore another security workflow
Turn security concerns into changes you can verify.
Start with a focused code excerpt, then review the proposed finding and fix with your existing tools.
Top up your AI balanceLLMsRelay provides model API access for authorized security research. Your tools run the tests; you verify the model's analysis.