Terms of Service
Please read these terms carefully before using the Latchkey AI-powered CI/CD monitoring and optimization services.
§01 Acceptance of Terms
By accessing or using the website located at https://latchkey.dev/ (the "Site") and the AI-powered CI/CD monitoring and optimization services provided by Latchkey, you agree to be bound by these Terms of Service.
If you do not agree to these Terms, do not use the Service.
Last Updated: November 20, 2026
Questions: legal@latchkey.dev
§02 Description of Service
Latchkey provides an analytics, runner, and optimization platform designed for engineering teams. Our Service integrates with GitHub Actions to provide predictive insights, cost visibility, and performance optimization recommendations, and offers managed ephemeral GitHub Actions runners with AI-driven self-healing of failing build steps.
Performance AnalyticsCost OptimizationAI-Powered InsightsManaged RunnersSelf-HealingAI Scan Runners
§03 Latchkey Runners
Latchkey Runners are managed, ephemeral GitHub Actions runners offered as part of the Service. You adopt a runner by setting the corresponding runs-on label (for example, latchkey-medium) in your workflow YAML. Latchkey provides four preset sizes (small, medium, large, extra-large) and AI Scan Runners that generate a runner configuration tailored to your repository.
3.1 Execution Environment
When a job is dispatched to a Latchkey Runner, your code is checked out into a fresh, isolated runner instance provisioned by Latchkey. The runner is destroyed at the end of the job, and nothing on it carries into another job. Apart from the job's logs and run records, Latchkey keeps a job's files in two cases. When you or your workflow ask it to, it keeps them in a cache entry a job saves, an artifact a job uploads to Latchkey or an output a job records; each is available only within your workspace and is deleted when the period our documentation states for it ends. When self-healing diagnoses a failing step, the heal record keeps what it read of the job's files for 90 days, and a fix it proposes keeps its file contents for 365 days. A working tree you upload from the Latchkey CLI to start a job is deleted after that job ends, apart from what self-healing read of it. Latchkey stores, retains and uses your source code only to execute your jobs, to keep what you ask it to keep, to diagnose and fix failing steps through self-healing, and for its other features that read your repositories, such as the knowledge base, repository scans and AI Scans, as our Privacy Policy describes and for the periods it states.
3.2 AI Scan Runners
AI Scan Runners analyze a monitored repository to detect languages, build tools, package managers, and CI configuration, and return a suggested runner configuration with a confidence score. The scan is a recommendation. You decide whether to approve, modify, or discard the generated configuration before any job is routed to it.
3.3 Your Responsibilities
You are responsible for the workloads you run on Latchkey Runners, including the code, dependencies, secrets, and third-party actions you choose to execute. You agree not to use Latchkey Runners for any purpose prohibited by these Terms (see "Prohibited Uses") or by GitHub's terms of service.
Latchkey Runners are fully compatible with existing GitHub Actions workflows. Standard GitHub Actions concepts (caching, artifacts, marketplace actions, secrets) continue to apply.
§04 Self-Healing
Self-healing is a capability of Latchkey Runners. When a workflow step fails, Latchkey detects the failure, diagnoses its cause, and applies a fix within a bounded action set so the build can keep moving. Self-healing targets transient and environmental failures - network errors, out-of-memory kills, disk-full errors, missing tools, and predictable configuration mismatches - and explicitly does not heal user-code bugs such as segfaults or deterministic crashes.
4.1 Default Behavior and Controls
Self-healing is enabled by default for every Latchkey workspace. Owners and admins can disable self-healing at any time under Settings → Self-healing. Other members can view the setting but not modify it.
4.2 Bounded Agent Actions
When the self-healing agent engages, it operates on a tightly scoped view of the failing step - stderr, stdout, and exit code - and can read files in the job's checkout or in the working tree you uploaded from the Latchkey CLI: at most four lookups for each diagnosis and the first 8 KiB of a file. Its available actions are limited to a vetted allowlist; it cannot run arbitrary commands, modify your code, or persist anything outside the build environment.
4.3 No Guarantee
Self-healing operates on a best-effort basis. We do not guarantee that any given failure will be successfully diagnosed or remediated, and you remain responsible for reviewing your build outcomes. Additional time consumed by self-healing is billed at the standard per-minute rate of the runner the failing step is using.
§05 Beta Services
Please Note: The Service is currently in an "Open Beta" or "Early Access" phase.
5.1 Beta Terms
- As-Is Status: The Beta Service is provided "as is" and "as available," and may contain bugs, errors, or other defects.
- Data Integrity: We do not guarantee that your data will be preserved without corruption or loss during the Beta phase.
- Feedback: You agree that we may use any feedback, suggestions, or ideas you provide without any obligation or compensation to you.
§06 Accounts and Integration
6.1 Registration
To use Latchkey, you must register for an account using one of the sign-in methods the sign-in page offers, which may include GitHub, Google, or a code sent to your email address. To use features that work with your repositories through your Git provider, you must also connect the Service to that provider (for example, by installing the Latchkey GitHub App).
6.2 Permissions
You grant Latchkey the necessary permissions to access, read, and analyze your CI/CD pipeline data, workflow YAML files, runner telemetry, the job files self-healing reads (including a working tree you upload from the Latchkey CLI), and the files of your repositories that Latchkey's other features read, as our Privacy Policy describes, solely for providing the Service. The Latchkey GitHub App asks for one set of permissions whichever features you use, including some that no Latchkey feature uses today; our documentation lists each permission and what it is for.
6.3 Pull Request Creation
The Latchkey GitHub App has write access to the repositories you choose to monitor so that Latchkey can open pull requests: the ones you start (for example, by applying an AI Insight finding) and heal pull requests, which self-heal opens on its own when it finds a lasting fix for a failure in a job that ran on Latchkey. Latchkey commits only to new branches it creates, never pushes to a branch of yours and never merges code. On Latchkey CI it also writes check runs on your commits. You review and merge any pull request Latchkey creates.
6.4 Runner Workloads
If you use Latchkey Runners, you direct your workflow jobs to Latchkey's infrastructure by setting the corresponding runs-on label in your workflow YAML. You remain solely responsible for the workloads you execute on Latchkey Runners and for ensuring those workloads comply with these Terms, GitHub's terms, and all applicable law.
6.5 Security
You are responsible for maintaining the confidentiality of your account credentials and for safeguarding any secrets, tokens, or credentials you provide to Latchkey Runners through GitHub Actions.
§07 Subscription and Billing
Latchkey offers tiered subscription plans sized by the number of monitored repositories you analyze and the number of seats included on the plan. Latchkey Runner usage is metered separately on a per-minute basis, with a pool of free runner minutes included with every plan.
When the Trial Ends
The trial needs no card. When it ends, runs stop until you add a payment method and choose a plan; you are charged when you subscribe.
7.1 Billing Terms
- Subscription tiers are sized by the number of monitored repositories included in your plan.
- Per-seat fees apply based on your selected plan. Each plan includes a baseline seat count, and additional seats can be purchased at the current per-seat rate.
- Latchkey Runner usage is metered separately on a per-minute basis. Each plan includes a pool of free runner minutes per calendar month, and usage beyond the included pool is billed at the published per-minute rate for the runner size in use.
- Runner minutes beyond the included pool are charged on their own invoice on the 1st of each month (UTC), for the month just ended.
- A plan you choose is charged when you subscribe.
- You can cancel at any time via your account dashboard to prevent future charges.
- When a cancelled plan ends, the runner minutes run in its last month are charged on the next 1st (UTC); when a workspace is deleted, its runner minutes for the month so far are charged on a final invoice.
- No refunds are provided for partial billing periods.
7.2 Payment Processing
All payments are securely processed through Stripe. We do not store your credit card information on our servers.
§08 AI Insights and Accuracy
Our Service uses AI in several places, including optimization recommendations surfaced on the dashboard, AI Scan runner configurations, and self-healing diagnoses and remediations applied to failing build steps.
No Guarantee
AI models can make mistakes. We do not guarantee specific results from following our recommendations, from any AI Scan runner configuration, or from any self-healing action. You are responsible for reviewing changes Latchkey proposes and for verifying that any healed build behaves correctly. Where AI actions involve cost (for example, self-healing minutes), those costs are billed at the standard per-minute rate of the runner in use.
§09 Intellectual Property
9.1 Our Rights
Latchkey retains all rights to the Service, including our AI models, algorithms, and underlying code.
9.2 Your Data
You retain ownership of your data. You grant us a limited license to use it to provide the Service, including the generation of account-specific insights.
9.3 AI and Machine Learning
To improve Latchkey for all users, we may use anonymized metadata (such as job durations and failure patterns) for model training, provided such data cannot be traced back to you or your workspace. We do not use your proprietary logs or source code to train global, multi-tenant AI models.
9.4 Opt-Out
Latchkey may use aggregated, non-identifiable data to improve its predictive models. Users on Scale or Enterprise tiers may request to opt-out of global model training by contacting legal@latchkey.dev.
§10 Comparative Statements and Trademarks
Latchkey publishes educational and comparative content, including guides, glossaries, and product comparisons in our Learn hub, that may reference third-party products, services, and companies.
10.1 Third-Party Trademarks
All third-party names, logos, product names, and trademarks referenced in our content are the property of their respective owners. Any such reference is for identification, commentary, and comparison purposes only and constitutes nominative fair use. It does not imply any affiliation with, sponsorship by, or endorsement by the trademark owner, and does not imply that Latchkey is endorsed by or affiliated with those owners.
10.2 Basis of Comparisons
Comparative statements are made in good faith and are based on information Latchkey believes to be accurate at the time of publication, drawn from publicly available sources such as each third party's own documentation, pricing pages, and public announcements, as of the date indicated on each page. Third-party features, pricing, availability, and terms change frequently and are outside Latchkey's control. Latchkey does not warrant that such information is current, complete, or error-free, and you should verify any third-party detail directly with the relevant provider before relying on it.
10.3 Opinions and No Reliance
Comparative content may reflect Latchkey's own views and opinions and is provided for general informational purposes only. It is not a representation, warranty, or guarantee regarding any third party or its products or services, and it is not a substitute for your own independent evaluation. Latchkey is not responsible for decisions made in reliance on this content.
Corrections
We aim to keep comparative content fair and accurate. If you are a third party or a reader and believe any statement is inaccurate or out of date, contact legal@latchkey.dev and we will promptly review and, where appropriate, correct or remove it.
§11 Prohibited Uses
You agree not to:
- Reverse engineer, decompile, or disassemble any aspect of the Service.
- Use the Service to benchmark without our prior written consent.
- Use the Service for any illegal purpose.
§12 Third-Party Services
The Service depends on third-party platforms (e.g., GitHub, GitLab).
Third-Party Disclaimer
We are not responsible for interruptions caused by third-party providers.
§13 Limitation of Liability
To the maximum extent permitted by law, Latchkey shall not be liable for any indirect, incidental, special, consequential, or punitive damages.
§14 Termination
We reserve the right to suspend or terminate your access to the Service at our sole discretion for conduct that violates these Terms.
§15 Contact Us
If you have questions about these Terms, please contact us at:
Changes to These Terms
We reserve the right to modify these Terms at any time. Your continued use constitutes acceptance of the new Terms.