Activities of "k.alamri"

Hi ABP team, We need help with two issues affecting the same development batch: Commercial licensing in an isolated integration-test host, and repeated Studio AI context-limit/timeout errors. We have reviewed the documentation, checked the samples index and searched existing support discussions. We still need a confirmed 10.7 licensing approach and practical guidance for continuing AI-assisted work without repeatedly rebuilding context and consuming additional credits.

1. Environment

• OS: Windows.
• ABP Framework and Commercial packages: 10.7.0.
• LeptonX: 5.7.0.
• Target framework: net10.0.
• Recorded .NET SDK: 10.0.401.
• Volo.Abp.Studio.Client.AspNetCore: 3.1.1.
• Running ABP Studio version: [INSERT VERSION].
• Application UI: React, with Commercial web modules in the backend.
• Test database: in-memory SQLite, not EF Core InMemory.
• Test host: ASP.NET Core TestServer using WebApplication.CreateEmptyBuilder, with EnvironmentName = "Development" explicitly set in code. The Commercial module graph remains enabled, including Account Pro, Identity Pro, OpenIddict Pro and SaaS.

2. Licensing failure: exception messages and diagnostic evidence The test project builds successfully, but the test host terminates during InitializeApplicationAsync, before any tests execute. The captured diagnostic contains: ABP-LIC-ERROR - License check failed for 'Volo.Abp.Account.Pro.Public.Web.Shared-v10.7.0.0'.

ABP-LIC-0020 - License code not found! Ensure that your appsettings.json or appsettings.secrets.json has "AbpLicenseCode" key with your license code. The diagnostic sequence reaches: CreateEmptyBuilder AddApplicationAsync Build InitializeApplicationAsync OnPreApplicationInitialization License errors Process exit The recorded host exit code was -42, with zero tests executed. We do not have a full exception stack from the licensing component. A temporary process-exit diagnostic captured only the exit callback, not the original licensing call stack. Please advise how to collect the appropriate diagnostic safely. Steps to reproduce in our fixture

1. Build the test project against the installed ABP Commercial 10.7.0 packages.
2. Create a Development host using WebApplication.CreateEmptyBuilder.
3. Supply only explicitly allowed, non-secret in-memory configuration.
4. Configure TestServer and in-memory SQLite while retaining the application’s Commercial modules.
5. Add the application module, build the host and call InitializeApplicationAsync.
6. Observe the licensing errors and process termination before test execution. Our isolated configuration intentionally excludes appsettings.secrets.json, unrelated environment variables and production connection strings. It currently has no AbpLicenseCode entry, so we understand why the missing-code error occurs. We propose supplying our legitimate license code through one dedicated in-memory configuration entry. We need to understand the remaining requirements before doing that. Licensing questions
7. Supported input: Is supplying IConfiguration["AbpLicenseCode"] through an in-memory provider supported in 10.7.0, without loading appsettings.secrets.json?
8. Offline versus online: Can this Development test host validate using that code alone? What determines whether online validation runs: the host environment, process environment variables, debugger state or another signal? Does setting Development through CreateEmptyBuilder behave like a standard host?
9. Developer authentication: If a login is required, does signing into ABP Studio suffice, or must the Windows account executing the tests also run abp login? Which credential files must the runtime access?
10. Network requirements: What destinations, ports and protocols are required? What information is transmitted during validation, including license, account, machine or application identifiers? Are proxies and TLS inspection supported?
11. Cache and local access: Is there a validation cache or offline grace period? Where is it stored, how long is it valid, and what files does validation read or write? Does validation happen per process or per host instance? Our suite creates several isolated hosts.
12. CI: What is the supported unattended authentication method for Commercial integration tests? Please clarify account ownership, credential renewal, ephemeral runners and any machine/concurrency limits.
13. Supported example: Could you provide a minimal 10.7 example or documentation for an isolated Commercial integration-test host? We are not trying to bypass licensing, remove Commercial modules or switch environments merely to avoid validation. Our fixture currently rejects outbound requests through its configured IHttpClientFactory pipeline. We understand this is not a process-wide network firewall and need your guidance on the actual licensing requirements.
14. Studio AI failures: exact messages During planning and implementation sessions, we repeatedly encountered: ⚠️ Error: The conversation is too large for the selected model's context window. Please start a new chat or select a model with a larger context window. We also encountered: ⚠️ Error: AI response timed out. The model took too long to respond. Please try again. And: ⚠️ Error: ABP Studio AI encountered an error: Unable to read data from the transport connection: An existing connection was forcibly closed by the remote host. These are the messages displayed by Studio. No full underlying exception stack was shown. Observed sequence
15. Continue a planning or implementation session that has accumulated source reads, plan content, diagnostics and documentation.
16. Send the next instruction.
17. Receive a context-window error; changing models did not reliably make the existing session usable.
18. Start a new session and reference saved work to continue.
19. Re-read and reconcile existing files because the previous error did not reliably establish which changes had been saved. A new session has also reported no active plan even though the plan file exists. We have preserved work on disk, but recovering its exact state has required additional effort and context. These are observed steps, not a minimal deterministic reproduction. Please tell us which session identifiers and diagnostics you need to investigate. Context recovery and credit questions What is ABP’s recommended workflow for avoiding these errors and continuing without losing work or unnecessarily consuming more credits? Specifically:

• Where should we configure the context limit and Max in our Studio version? How does Studio account for the selected model’s actual limit, tool output and reserved response tokens?
• Does Studio support automatic context compaction, durable summaries, memory or checkpoints? What persists across sessions, and how should we resume an existing saved plan without replaying completed work?
• After a timeout or connection failure, how can we determine whether a tool operation completed or a file was saved before retrying?
• Are requests rejected for excessive context charged? How are timeouts, transport failures and automatic retries billed?
• Can you review the affected requests and restore/refund eligible credits consumed by failed or duplicate attempts? Please explain the policy and what evidence we should provide. We are not assuming every failed request was charged or that all previous work was lost. Please also explain how to export the relevant session IDs, request IDs, timestamps and usage records safely.

4. References already checked

• ABP AI Agent configuration — including context limits and Max.
• ABP samples.
• Licensing support discussion #10636.
• Development-container licensing discussion #5856. The older licensing discussions do not establish the exact behavior of our 10.7 setup. Please have the appropriate engineering and billing teams review these questions. If separate tickets are required, please let us know how to split and cross-reference them. We can provide sanitized screenshots, diagnostic excerpts and a host-construction snippet. We will not attach license codes, access tokens, secrets files or credential-bearing feed configuration. Thank you.