* refactor: update headerList methods and translations for consistency
- Renamed methods in req.headerList and res.headerList from 'forEach' to 'each' for consistency with the new API.
- Updated method translations in the Postman converters to reflect the new method names: 'append' to 'add', 'set' to 'upsert', and 'delete' to 'remove'.
- Adjusted related tests to ensure they validate the new method names and functionality.
- Removed deprecated test cases for 'append' and 'set', replacing them with tests for 'add' and 'upsert'.
- Enhanced documentation to clarify the changes in method names and their usage.
* test: add new tests for HeaderList methods and behavior
- Introduced tests to verify that the 'idx' property is undefined in HeaderList, ensuring compliance with the updated API.
- Added tests to confirm that positional methods (prepend, insert, insertAfter) do not exist in HeaderList, reflecting the recent refactor.
- Implemented a test to check that the two-argument form of the 'add' method correctly overwrites existing headers, enhancing the robustness of header management tests.
* refactor: remove 'idx' method from headerList and update related tests
- Eliminated the 'idx' method from both req.headerList and res.headerList to streamline header management.
- Updated associated tests and documentation to reflect the removal, ensuring clarity in the API usage and maintaining consistency across the header management system.
* refactor: block unimplemented HeaderList methods with error messages
- Added error handling for unimplemented methods in HeaderList, including idx, add, upsert, remove, each, prepend, insert, and insertAfter.
- Each method now throws a descriptive error indicating the appropriate alternative methods to use, enhancing clarity in the API and guiding users towards correct usage.
* refactor: update error message in idx() method of HeaderList for clarity
- Modified the error message in the idx() method to guide users towards using all()[index] or get(name) instead of the unsupported idx() method, enhancing the clarity of the API documentation.
* refactor: update HeaderList to extend ReadOnlyPropertyList and remove unimplemented methods
- Changed HeaderList to extend ReadOnlyPropertyList instead of PropertyList, streamlining its functionality.
- Removed unimplemented methods (prepend, insert, insertAfter) from HeaderList, clarifying the API and guiding users towards using supported methods.
- Updated related tests to reflect these changes, ensuring consistency and accuracy in header management.
* test: add translations for pm.request.headers methods in request tests
- Introduced new tests to validate the translation of pm.request.headers methods (prepend, insert, insertAfter) to their corresponding req.headerList.append method.
- Enhanced existing tests to ensure accurate conversion and functionality of header management in the Bruno converters.
* feat: enhance header translation methods for pm.request.headers
- Added translations for additional pm.request.headers methods (get, has, one, all, count, indexOf, find, filter, each, map, reduce, toObject, clear) to their corresponding req.headerList methods.
- Updated tests to validate the new translations and ensure accurate header management functionality in the Bruno converters.
* feat(snapshot): add session snapshot persistence and restoration
- Add snapshot middleware to persist UI state (tabs, workspaces, environments)
- Add SnapshotManager service in electron for atomic snapshot storage
- Add accessor-based tab serialization using pathname for reliable restoration
- Add loading states for tabs while collections are mounting
- Add hydrateTabs to restore tabs from snapshots on app load
- Add devTools state persistence (console open/height/tab)
* fix(snapshot): preserve unloaded collections and fix async serialization
Make serializeSnapshot async to fetch existing snapshot before saving,
ensuring collections not currently loaded in Redux are preserved. Fix
activeTab serialization to pass collection object instead of just UID.
* refactor(snapshot): rewrite storage to map-based schema with granular IPC
Replace array-based snapshot storage with key-value maps keyed by pathname
for O(1) lookups. Separate tabs into their own top-level map, decoupled
from collections.
- Rewrite SnapshotManager with map-based schema and granular read/write
methods (getWorkspace, getTabs, setCollection, removeWorkspace, etc.)
- Add 12 granular IPC handlers (renderer:snapshot:*) replacing 4 coarse ones
- Update middleware serialization to produce maps; remove activeCollectionUidCache
in favor of lastActiveCollectionPathname on workspace objects
- Fix mountCollection passing collectionUid instead of collection to restoreTabs
- Preserve non-active workspace state from existing snapshot on save
* wip
* refactor: allow migration of old snapshot
* refactor: trim down redundancy
* fix: for workspace state
* feat: fix for finalised schema
* fix: schema cleanup
* chore: simplify
* chore: wait on hydration to finish
* chore: snapshot changes to schema
* chore: fix typo
* fix: switch schema for saving and restoring extras
* chore: correctness changes
* chore: add active in the schema check
* chore: fix lint
* chore: comments
* chore: make writes async
* fix: sorting and cross workspace shared collection fixes
* chore: add version
* fix: wait on hydration
* chore: fix backward compat for ui-snapshots
* chore: dead code removal
* Fix optional chaining in snapshot lookup
---------
Co-authored-by: Chirag Chandrashekhar <cchirag85@gmail.com>
* feat: introduce HeaderList for dynamic header management in BrunoRequest and BrunoResponse
- Implemented HeaderList to provide a dynamic API for managing request headers, allowing real-time reflection of changes.
- Updated BrunoRequest and BrunoResponse to utilize HeaderList for header access, enhancing consistency and usability.
- Added a headers proxy to maintain backward compatibility with existing header access patterns.
- Introduced tests for HeaderList to ensure functionality and reliability across various header operations.
* refactor: update createHeadersProxy to accept a getter function for raw headers
- Modified createHeadersProxy to allow passing a function that retrieves raw headers, enhancing flexibility in header management.
- Updated HeaderList to utilize the new getter function, ensuring consistent behavior when headers are modified.
- Added a test to verify that bracket access reflects changes made via BrunoRequest.setHeaders.
* refactor: enhance headers proxy and syncWriteMethods for improved header management
- Updated createHeadersProxy to clarify precedence of PropertyList methods over header names in bracket access.
- Expanded syncWriteMethods in bruno-request shim to include additional methods for better header manipulation.
- Implemented a custom remove method in property-list-bridge to handle function predicates in-VM, improving the native bridge's functionality.
* refactor: update header management in BrunoRequest and BrunoResponse
- Renamed `req.headers` to `req.headerList` for improved clarity and consistency in header operations.
- Enhanced `HeaderList` to support both dynamic and static modes for header management.
- Updated shims for `req` and `res` to reflect the new header access patterns.
- Modified tests to validate the new header access methods and ensure backward compatibility with raw headers.
* refactor: enhance header translation methods for BrunoRequest and BrunoResponse
- Added comprehensive translations for `req.headerList` and `res.headerList` methods to their respective Postman equivalents, improving consistency in header management.
- Updated tests to validate the new translation methods, ensuring accurate conversion between Bruno and Postman header operations.
- Enhanced existing tests to cover additional header manipulation scenarios, reinforcing the reliability of the header management system.
* refactor: streamline headerList implementation in BrunoRequest and BrunoResponse
- Replaced lazy initialization of `headerList` with direct instantiation in both classes for improved performance and clarity.
- Removed redundant getter methods for `headerList`, simplifying the API.
- Updated tests to reflect changes in headerList instantiation and ensure proper functionality.
* refactor: remove header proxy
* feat: add comprehensive headerList tests for request and response
- Introduced multiple test files for `req.headerList` and `res.headerList` methods, covering various operations such as add, remove, clear, populate, and search methods.
- Implemented tests for both dynamic and read-only scenarios, ensuring robust validation of header management functionalities.
- Enhanced existing headerList methods with detailed assertions to improve reliability and maintainability of the header management system.
* feat: expand STATIC_API_HINTS with additional headerList methods for req and res
- Added a comprehensive list of methods for `req.headerList` and `res.headerList` to enhance autocomplete functionality.
- Included various operations such as get, add, remove, and manipulation methods to improve developer experience and usability.
* feat: implement support for disabled headers in headerList
- Added functionality to track disabled headers in `req.headerList` and `res.headerList`, allowing for better management of header states.
- Updated the `prepareRequest` and `prepareGrpcRequest` functions to include a `disabledHeaders` property.
- Enhanced the `HeaderList` class to handle disabled headers, including methods to filter, count, and retrieve them.
- Introduced tests to validate the behavior of disabled headers, ensuring they are correctly included in the header list while being excluded from raw headers.
* feat: enhance HeaderList with case-insensitive key lookups and improved toObject method
- Implemented case-insensitive key lookups for methods such as get, one, has, and indexOf in HeaderList.
- Updated toObject method to support options for excluding disabled headers, handling duplicate keys, and skipping headers with falsy keys.
- Modified toString method to return headers in HTTP wire format, improving consistency with standard practices.
- Added comprehensive tests to validate new functionalities, ensuring robust header management and accurate behavior for both enabled and disabled headers.
* feat: add context binding to iteration methods in HeaderList
- Enhanced iteration methods (each, filter, find, map, reduce) in HeaderList to accept an optional context parameter, allowing for better control over the `this` binding in callback functions.
- Updated the remove method to support context binding for function predicates.
- Added comprehensive tests to validate the new context binding functionality across various iteration methods, ensuring robust header management.
* feat: enhance HeaderList with new methods for string handling and object support
- Added support for accepting an object with a key property in the `has` method, improving header existence checks.
- Updated `add` and `populate` methods to accept "Key: Value" strings and multi-line header strings, enhancing flexibility in header management.
- Modified `toString` method to ensure a trailing newline in the output, aligning with HTTP wire format standards.
- Introduced comprehensive tests to validate new functionalities, ensuring robust header management and accurate behavior for various input types.
* feat: enhance header management with support for disabled headers and context binding
- Updated `mergeHeaders` function to preserve disabled headers from the request item, improving header state management.
- Modified `addBrunoRequestShimToContext` and `addBrunoResponseShimToContext` to wrap eval code in blocks, preventing redeclaration conflicts.
- Enhanced iteration methods in `property-list-bridge` to support context binding, allowing for better control over `this` in callbacks.
- Added comprehensive tests for `req.headerList` and `res.headerList` to validate new functionalities and ensure robust header management.
* feat: improve header merging to preserve disabled headers from request items
- Enhanced the `mergeHeaders` function to retain disabled headers from the last entry in the request tree path, ensuring better management of header states.
- Updated the logic to include disabled headers while merging, improving the overall functionality of header management.
* feat: update header merging logic to consistently handle disabled headers
- Refactored the `mergeHeaders` function to utilize a separate array for disabled headers, ensuring they are preserved during the merging process.
- Simplified the logic for merging headers by directly combining enabled and disabled headers into the request object, enhancing clarity and maintainability.
* refactor: update disabled headers handling in mergeHeaders function
- Changed the implementation of disabled headers from an array to a Map for better performance and consistency.
- Updated the logic in the `mergeHeaders` function to ensure disabled headers are correctly merged into the request object.
- Enhanced tests to validate the handling of disabled headers, including new scenarios for folder-level and request-level overrides.
* feat: enhance mergeHeaders function to support optional inclusion of disabled headers
- Updated the `mergeHeaders` function to accept an options parameter, allowing for the inclusion of disabled headers in the merged result.
- Modified the logic to handle disabled headers more effectively, ensuring they can be returned based on the provided options.
- Adjusted calls to `mergeHeaders` in various modules to utilize the new functionality, improving header management consistency across the application.
* refactor: remove disabledHeaders from prepareGrpcRequest function
- Eliminated the handling of disabledHeaders in the prepareGrpcRequest function to streamline header management.
- Updated the headers processing logic to focus solely on enabled headers, enhancing clarity and reducing complexity.
* feat: add translations for req.headerList and res.headerList methods
- Introduced new tests to translate methods from req.headerList and res.headerList to their corresponding pm.request.headers and pm.response.headers methods.
- Added translations for methods: one, find, toObject, and upsert, enhancing the functionality of the header management system.
* docs: update HeaderList method aliases and add clarification notes
- Enhanced documentation for `prepend`, `append`, `insert`, and `insertAfter` methods to clarify that they are aliases for `add()` and include a note on the lack of support for ordering and duplicates.
- Updated method comments to reflect the internal handling of headers as a plain object, ensuring better understanding of header management behavior.
* fix: remove duplicate headerList initialization in BrunoRequest class
- Eliminated redundant initialization of headerList in the BrunoRequest constructor, ensuring cleaner code and preventing potential issues with header management.
- This change simplifies the constructor logic while maintaining the intended functionality of the headerList.
* refactor: remove unimplemented headerList methods and update documentation
- Removed the unimplemented methods `prepend`, `append`, `insert`, and `insertAfter` from the headerList, ensuring clarity in the API.
- Updated documentation to reflect that these methods are not implemented and provide guidance to use `add()` or `upsert()` instead.
- Adjusted tests to verify that calls to these methods throw appropriate not-implemented errors, enhancing the robustness of header management.
* refactor: remove unused headerList append method from pre-request script
- Eliminated the call to `req.headerList.append()` in the pre-request script, streamlining the header management process.
- This change aligns with the recent refactor to remove unimplemented methods, ensuring clarity and consistency in the API usage.
* test: update error messages for unimplemented headerList methods
- Changed error messages in tests for `append`, `prepend`, `insert`, and `insertAfter` methods to indicate they are "not yet implemented" instead of "not implemented".
- This update improves clarity in the test outputs, aligning with the current state of the headerList API.
* refactor: remove unimplemented headerList methods and update related tests
- Eliminated the unimplemented methods `prepend`, `append`, `insert`, and `insertAfter` from the HeaderList class and corresponding tests, clarifying the API usage.
- Updated the documentation to guide users towards using `add()` or `upsert()` instead.
- Adjusted tests to remove references to these methods, ensuring they accurately reflect the current state of the headerList API.
* refactor: update HeaderList to accept raw request and response objects
- Modified the HeaderList class to accept the raw request and response objects directly, simplifying the initialization process.
- Updated the BrunoRequest and BrunoResponse classes to reflect this change, ensuring consistent handling of headers.
- Enhanced internal methods to utilize the new structure, improving clarity and maintainability of the header management system.
* refactor: enhance HeaderList header management and update tests
- Updated HeaderList to track removed headers for axios interceptor, improving header casing handling.
- Added new syncWriteMethods to the BrunoResponse shim for better integration.
- Enhanced tests to verify the correct tracking of deleted headers and updated expectations for header management functionality.
* refactor: update header translations and improve HeaderList methods
- Removed unused 'idx' translations from both bruno-to-postman and postman-to-bruno translators to streamline header management.
- Enhanced the HeaderList class to skip existing keys when populating headers, ensuring that current values are preserved.
- Updated tests to reflect the new behavior of the populate method, verifying that existing headers are not overwritten and adjusting expectations accordingly.
* refactor: update headerList.populate behavior and adjust tests
- Modified the req.headerList.populate method to add new headers while preserving existing ones, ensuring that current values are not overwritten.
- Updated the associated test to reflect this new behavior, verifying that existing headers remain intact and new headers are correctly added.
* refactor: update HeaderList methods and translations for consistency
- Renamed `each` method to `forEach` in HeaderList for consistency with standard JavaScript array methods.
- Updated header management methods: replaced `add` with `append`, `upsert` with `set`, and `remove` with `delete` to better reflect their functionality.
- Enhanced translations in bruno-to-postman and postman-to-bruno converters to align with the new method names.
- Adjusted tests to verify the new method names and ensure correct header management behavior.
* refactor: remove unused headerList methods and update related shims
- Removed the `entries`, `keys`, and `values` methods from the HeaderList class to streamline the API and eliminate unused functionality.
- Updated the `bruno-request` and `bruno-response` shims to reflect the removal of these methods, ensuring consistency in header management.
- Adjusted tests to remove references to the deleted methods, maintaining alignment with the current state of the headerList API.
* fix: ensure re-added headers are removed from __headersToDelete
- Updated the HeaderList class to remove headers from the __headersToDelete array when they are re-added, preventing incorrect header deletion behavior.
- Added tests to verify that re-added headers do not remain in __headersToDelete and that their values are correctly set in the raw request headers.
* Add drag-resize split pane for API Spec viewer
Introduce a drag-to-resize split pane for the API Spec viewer and persist left pane width. Adds a new useDragResize hook to manage dragging state and clamping, plus UI: dragbar styles, a loading state for the Swagger preview (onComplete + loader), and memoization of the Swagger renderer. Wire up persisted widths via Redux: add updateApiSpecPanelLeftPaneWidth (apiSpec slice) and updateApiSpecTabLeftPaneWidth (tabs slice), and propagate leftPaneWidth / onLeftPaneWidthChange through ApiSpecPanel, OpenAPISpecTab, RequestTabPanel and SpecViewer. Misc: pass tab uid into OpenAPISpecTab and add .gstack/ to .gitignore.
* Refactor SpecViewer and OpenAPISpecTab for improved loading and state management
- Updated SpecViewer to enhance loading state handling for Swagger content, ensuring a smoother user experience by preventing flashes of unrendered content.
- Refactored OpenAPISpecTab to streamline environment context management, optimizing the loading process for OpenAPI specifications.
- Simplified the useDragResize hook by removing unnecessary references and improving the handling of drag events, ensuring better performance and responsiveness during resizing actions.
* Enhance useDragResize hook to clamp width seed and improve test coverage
- Updated the useDragResize hook to clamp the width seed value, ensuring it stays within defined bounds during drag events.
- Added a new test case to verify that an out-of-bounds width seed is correctly clamped and persisted on immediate mouseup, enhancing the robustness of the drag-resize functionality.
* Remove .gstack/ from .gitignore
Delete the .gstack/ ignore entry and normalize the packages/bruno-converters/dist entry in .gitignore (deduplicated). No code changes; just tidy up ignore rules.
* feat: implement side-by-side diff viewer for spec synchronization
- Added a new SpecDiffModal component to display differences between current and updated specs.
- Introduced buildRows function to flatten parsed diff data for rendering.
- Created DiffRow component for rendering individual rows in the diff view.
- Implemented highlightCache for efficient word-level diff highlighting.
- Enhanced user experience with navigation controls for changes and loading indicators.
- Added tests for buildRows functionality to ensure accurate diff representation.
* fix: update comments and dependencies for consistency in SpecDiffModal and StyledWrapper
- Added a comment in StyledWrapper.js to clarify the min-height requirement for Virtuoso's fixedItemHeight.
- Updated comment in highlightCache.js to reflect the change from character-level to word-level diff highlighting.
- Adjusted dependency array in SpecDiffModal.js to include cache for improved performance.
* Centralize OpenAPI sync state in Redux
Move per-collection OpenAPI sync state into the Redux slice and update callers. Adds storedSpec and drift maps and reducers (setDrift, clearDrift, setStoredSpec, clearOpenApiSyncTabState), removes the old diff payload from collectionUpdates, and keeps storedSpecMeta. Components and hooks were updated to use inline selectors (state.openapiSync...) instead of exported selectors, and useOpenAPISync was refactored to persist drift/storedSpec to Redux, use the store to read tabs, and call setDrift/setStoredSpec rather than keeping duplicate local state. The collections closeTabs action now clears openapi-sync state for closing openapi-sync tabs so transient state is dropped when tabs are closed. Small variable renames and minor logic adjustments to use the new shape were included.
* Enhance useOpenAPISync hook with comments for clarity on store usage. This update clarifies that tabs are read-only within handlers to prevent unnecessary re-renders when using useSelector.
* fix: collection settings access, UI overflow fixes, and auto-focus URL bar
- Move collection icon outside dropdown; clicking it opens collection
settings/overview, clicking the name opens the switcher dropdown
- Auto-focus URL bar when creating a transient request (#2919)
- Fix long collection/folder name overflow with ellipsis truncation
- Reduce dropdown width and truncate large collection names
- Simplify breadcrumb collapse: show collection name and last folder,
collapse middle items into a dropdown
- Fix modal width to prevent shrinking with short collection names
- Show "Create Collection" option when saving a draft with zero collections
- Use IconBox consistently for collection icons
* Replace Chevron component with IconChevronRight
---------
Co-authored-by: Chirag Chandrashekhar <cchirag85@gmail.com>
Co-authored-by: Sid <siddharth@usebruno.com>
* fix: avoid round trip loss of annotation data
* feat: update types for file , multipart and tests for the same
* chore: optional
* chore: fix body:file annotation
* chore: remove log
* Add proxy .pac file resolver
chore(dependencies): update package-lock.json with new dependencies and version upgrades
- Added new dependencies: ajv, git-url-parse, @opencollection/types, and storybook packages.
- Updated existing dependencies to their latest versions, including eslint and babel packages.
- Removed deprecated entries and cleaned up the package-lock structure for better maintainability.
* tests
* wip
* wip
* wip
* wip
* feat: file upload .pac
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* feat: Refactor proxy settings to use a new structure for PAC configuration. Introduced 'source' field to determine proxy type (manual or PAC) and updated related validation and state management. Removed deprecated 'pacUrl' field in favor of 'pac.source'. Updated preferences schema and test data accordingly.
* fix: Update proxy settings to correctly reference 'source' field for PAC configuration. Adjusted state management and validation logic to align with new structure. Enhanced tests for backward compatibility and new format handling.
* feat: Enhance proxy configuration by adding 'proxyModeReason' to provide context for proxy settings. Updated related functions to accommodate the new parameter and improved logging for proxy mode changes.
* wip
* refactor: Update proxy settings to remove 'inherit' field and replace it with 'source' for better clarity. Adjusted validation schema, default preferences, and migration logic to align with the new structure. Enhanced tests to ensure compatibility with the updated proxy configuration.
* wip
* wip
* wip
* wip
* wip
* chore: consistent path check
* chore: consistency
* tests(pac): fix unit params
---------
Co-authored-by: Gianluca D'Abrosca <gianluca.dabrosca.1999@gmail.com>
Co-authored-by: Sid <siddharth@usebruno.com>
* feat: enhance environment variable resolution in EnvironmentVariablesTable
- Added logic to populate process environment variables from the active workspace when no collection is selected, allowing for proper resolution of {{process.env.X}}.
- Updated workspace actions to map scratch collections to their respective workspaces, ensuring that environment variables can be resolved correctly.
This improves the user experience by providing access to workspace-specific environment variables in the environment variables table.
* refactor: improve state selection and add test IDs for better testing
- Refactored the state selection logic in EnvironmentVariablesTable for clarity.
- Added data-testid attributes to various components including CollapsibleSection, DotEnvFileDetails, DotEnvRawView, and EnvironmentList to enhance testability.
- This change aims to streamline component interactions and facilitate easier testing.
* feat: add test IDs to EnvironmentList for improved testability
- Introduced data-testid attributes to the EnvironmentList component, enhancing the ability to target elements in tests.
- This update includes test IDs for the CollapsibleSection, create .env file button, and the input field for the .env name, facilitating better integration with testing frameworks.
* refactor: simplify global environment test setup
- Removed unnecessary timeout setting and afterEach cleanup logic from the global environment process.env resolution test.
- This change streamlines the test structure, focusing on the core functionality being tested.
* fix: update system proxy fetching to use finally (#7652)
* fix: update system proxy fetching to use finally for improved reliability
* Update packages/bruno-electron/src/index.js
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* fix: allow file selection in multipart form without entering a key first (#7640)
* fix: close previous SSE connection before sending new request
When resending an SSE (Server-Sent Events) request using Cmd+Enter,
the previous connection was not being closed, causing connection leaks.
Changes:
- Add SSE cancellation logic to sendRequest action - checks for running
stream and cancels it before sending new request
- Add return to cancelRequest action to make it properly chainable
- Simplify RequestTabPanel by removing duplicate cancel logic (now
handled centrally in sendRequest)
- Add SSE endpoints to test server for e2e testing
- Add Playwright e2e test to verify SSE connection cancellation
* fix: address PR review feedback for SSE connection cancellation
- Use platform-aware modifier (Meta on macOS, Control on Linux/Windows)
instead of hardcoded Meta+Enter for cross-platform CI compatibility
- Replace waitForTimeout with expect.poll for deterministic assertions
- Remove dead try/catch around cancelRequest (errors already swallowed
by cancelRequest's internal .catch)
* fix: updated the test to check of connectionIds
---------
Co-authored-by: Sid <siddharth@usebruno.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Pooja <pooja@usebruno.com>
Co-authored-by: Chirag Chandrashekhar <cchirag85@gmail.com>
* feat: add gRPC proxy support
* fix: respect channelOptions grpc.primary_user_agent over User-Agent header
* fix: remove non-standard grpc_proxy/no_grpc_proxy env var support
These are not recognized by any standard gRPC implementation. gRPC proxy
now uses the standard http_proxy/https_proxy/no_proxy variables like
grpc-core, grpc-go, and grpc-java.
* chore: add resolveGrpcProxyConfig tests and clean up grpc-client
Export resolveGrpcProxyConfig for testability and add unit tests covering
all proxy modes (off, on, system), auth encoding, protocol rejection,
bypass lists, and edge cases. Remove redundant cancelAndCloseConnection
call in startConnection (already guarded by addConnection). Document why
internal @grpc/grpc-js channel options are used for programmatic proxy.
