Repository navigation
refactor: unify call type and ensure typesafety - #2659
Merged
Merged
Conversation
Lightning00Blade
force-pushed
the
pre-fix-types
branch
4 times, most recently
from
September 17, 2026 12:44
1401270 to
80dc15b
Compare
Lightning00Blade
marked this pull request as ready for review
September 17, 2026 12:44
Lightning00Blade
force-pushed
the
pre-fix-types
branch
from
September 17, 2026 14:09
80dc15b to
4386f5a
Compare
OrKoN
reviewed
Sep 18, 2026
OrKoN
reviewed
Sep 18, 2026
Lightning00Blade
force-pushed
the
pre-fix-types
branch
from
September 18, 2026 11:09
45a56e2 to
ee12b8e
Compare
Lightning00Blade
force-pushed
the
pre-fix-types
branch
from
September 18, 2026 11:38
8591b5f to
69e21b7
Compare
OrKoN
approved these changes
Sep 18, 2026
Lightning00Blade
enabled auto-merge
September 18, 2026 12:01
pull Bot
pushed a commit
to oogalieboogalie/chrome-devtools-mcp
that referenced
this pull request
Sep 23, 2026
🤖 I have created a release *beep* *boop* --- ## [1.10.0](ChromeDevTools/chrome-devtools-mcp@chrome-devtools-mcp-v1.9.0...chrome-devtools-mcp-v1.10.0) (2026-09-23) ### 🎉 Features * add css formatter class ([ChromeDevTools#2707](ChromeDevTools#2707)) ([f0b5fa4](ChromeDevTools@f0b5fa4)) * add get_css_style tool ([ChromeDevTools#2612](ChromeDevTools#2612)) ([4454ae0](ChromeDevTools@4454ae0)) * **performance:** chunked trace buffer parser for large recordings ([ChromeDevTools#2721](ChromeDevTools#2721)) ([23b9a48](ChromeDevTools@23b9a48)) * set default pageSize and pageIdx for get_css_styles tool ([ChromeDevTools#2799](ChromeDevTools#2799)) ([dc9d14d](ChromeDevTools@dc9d14d)) * support config file ([ChromeDevTools#2661](ChromeDevTools#2661)) ([314a5fa](ChromeDevTools@314a5fa)) * **telemetry:** persist date of the last tool call. ([ChromeDevTools#2705](ChromeDevTools#2705)) ([c54a493](ChromeDevTools@c54a493)) * **telemetry:** recording a sanitized version of the client name ([ChromeDevTools#2757](ChromeDevTools#2757)) ([55fbc57](ChromeDevTools@55fbc57)) * **telemetry:** report hermes client usage. ([ChromeDevTools#2703](ChromeDevTools#2703)) ([fb47e6c](ChromeDevTools@fb47e6c)) * update css formatter class to add AtRule, PositionTryRule, PropertyRule, FunctionRule ([ChromeDevTools#2717](ChromeDevTools#2717)) ([d4a0620](ChromeDevTools@d4a0620)) * update css formatter class to add inherited rules ([ChromeDevTools#2715](ChromeDevTools#2715)) ([3d7e7bb](ChromeDevTools@3d7e7bb)) * update css formatter class to add matched rules ([ChromeDevTools#2713](ChromeDevTools#2713)) ([8c8616f](ChromeDevTools@8c8616f)) * update css formatter class to add pseudo element ([ChromeDevTools#2716](ChromeDevTools#2716)) ([df1469a](ChromeDevTools@df1469a)) * update css formatter to add keyframes rules ([ChromeDevTools#2718](ChromeDevTools#2718)) ([cdc365c](ChromeDevTools@cdc365c)) ### 🛠️ Fixes * bound ConsoleCollector retention per navigation ([ChromeDevTools#2773](ChromeDevTools#2773)) ([e98a3ca](ChromeDevTools@e98a3ca)) * **cli:** forward explicit false options on start ([ChromeDevTools#2702](ChromeDevTools#2702)) ([d9a8cb6](ChromeDevTools@d9a8cb6)) * **config:** preserve raw values for config coercion ([ChromeDevTools#2747](ChromeDevTools#2747)) ([906c83b](ChromeDevTools@906c83b)) * don't log Puppeteer logs to file unless requested ([ChromeDevTools#2743](ChromeDevTools#2743)) ([4fbfbc4](ChromeDevTools@4fbfbc4)) * explain launch failures caused by running as root ([ChromeDevTools#2634](ChromeDevTools#2634)) ([9d29223](ChromeDevTools@9d29223)) * handle JavaScript dialogs opened during input tool actions ([ChromeDevTools#2794](ChromeDevTools#2794)) ([266112b](ChromeDevTools@266112b)) * **performance:** prevent memory leak by scoping trace engine model per parse ([ChromeDevTools#2720](ChromeDevTools#2720)) ([d05cbc0](ChromeDevTools@d05cbc0)) * preserve console history across same-document navigations ([ChromeDevTools#2676](ChromeDevTools#2676)) ([aa25562](ChromeDevTools@aa25562)) * preserve underlying error message in input tool actions ([ChromeDevTools#2792](ChromeDevTools#2792)) ([6e47dbb](ChromeDevTools@6e47dbb)) * set emulatedUserAgent and use finalDisplayedUrl in lighthouse_audit ([ChromeDevTools#2795](ChromeDevTools#2795)) ([941f82a](ChromeDevTools@941f82a)) * timout in WaitForHelper.ts ([ChromeDevTools#2772](ChromeDevTools#2772)) ([dc1d055](ChromeDevTools@dc1d055)) ### 📄 Documentation * add FLUJO client configuration ([ChromeDevTools#2690](ChromeDevTools#2690)) ([65a679e](ChromeDevTools@65a679e)) * fix categories not being configured correclty ([ChromeDevTools#2807](ChromeDevTools#2807)) ([2250cdc](ChromeDevTools@2250cdc)) * fix defaults for non-boolean values ([ChromeDevTools#2744](ChromeDevTools#2744)) ([342d243](ChromeDevTools@342d243)) * update page routing option name ([ChromeDevTools#2748](ChromeDevTools#2748)) ([8939965](ChromeDevTools@8939965)) * update SKILL for chrome-devtools to use get_css_styles tool ([ChromeDevTools#2760](ChromeDevTools#2760)) ([421011e](ChromeDevTools@421011e)) ### 🏗️ Refactor * disable tools instead of not registering them ([ChromeDevTools#2636](ChromeDevTools#2636)) ([b455469](ChromeDevTools@b455469)) * extract browser in a BrowserManager class ([ChromeDevTools#2787](ChromeDevTools#2787)) ([3228f44](ChromeDevTools@3228f44)) * migrate the MCP SDK to v2 ([ChromeDevTools#2408](ChromeDevTools#2408)) ([da3c406](ChromeDevTools@da3c406)) * move comments formatting to CommentFormatter class ([ChromeDevTools#2719](ChromeDevTools#2719)) ([e3cded0](ChromeDevTools@e3cded0)) * prepare for SDK v2 ([ChromeDevTools#2771](ChromeDevTools#2771)) ([61780c7](ChromeDevTools@61780c7)) * simplify our gen scritps ([ChromeDevTools#2793](ChromeDevTools#2793)) ([5068584](ChromeDevTools@5068584)) * unify call type and ensure typesafety ([ChromeDevTools#2659](ChromeDevTools#2659)) ([882f93e](ChromeDevTools@882f93e)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please).
bitbay
added a commit
to bitbay/chrome-devtools-mcp
that referenced
this pull request
Sep 23, 2026
…connection Re-implemented as a fresh diff against upstream's current ToolHandler.ts (post-ChromeDevTools#2659/ChromeDevTools#2787, which predates this branch's whole timeout-wrapping fix), not a literal merge of this branch's prior diff — unrelated upstream divergences (.strict() input schema, `disabled` field, `handle` as a class-field arrow function using `using _guard = ...`) are left as-is. Replaces the ToolHandler.ts portion of this branch's 6d2d4e9, and f67b230 in full: - TOOL_CALL_TIMEOUT_MS, ToolCallTimeoutError, and #raceWithTimeout<T>() ported verbatim. Puppeteer gives no way to cancel a pending call, so a timeout only stops the handler from waiting — the underlying attempt keeps running abandoned, which is exactly why the onTimeout callbacks below exist. - getContext() is wrapped in #raceWithTimeout, calling the injected abandonPendingBrowserAttemptOnTimeout on timeout (see the BrowserManager commit) so a resolution nobody is waiting for anymore doesn't get silently installed for a later caller. - The tool.handler() + response.handle() sequence is wrapped in a single shared budget (matching f67b230's reasoning: several tools' actual CDP calls happen inside response.handle(), not tool.handler(), so both need covering by one timeout or the second phase hangs unprotected after the first one's already accounted for). On timeout, forgetBrowserOnTimeout is called with the resolved context's browser. - src/index.ts: threads BrowserManager's forget()/abandonPendingAttempt() down as bound closures into ToolHandler's two new required constructor params (`forgetBrowserOnTimeout`, `abandonPendingBrowserAttemptOnTimeout`) instead of the old free-function defaults browser.ts used to provide, since BrowserManager is now private instance state owned by McpServer rather than importable module state. #getContext() itself needs no change — it already calls this.#browserManager.ensureBrowser(). This branch's 3d9dd88 (AGENTS.md style fixes: no `!`, no hand-rolled mocks, sinon.assert over Node's assert) has nothing to port forward — this diff was written directly against those rules and upstream's current tests/mocks.ts conventions, so there was nothing to retrofit. Tests: ~23 existing `new ToolHandler(...)` call sites updated for the two new required constructor params (no-op sinon.spy() stand-ins where the test doesn't care about them). Three new regression tests ported from this branch's prior versions: a hung tool.handler() times out and forgets the browser; a hung response.handle() times out even when tool.handler() resolves fast; a hung getContext() times out and abandons the pending attempt without touching forgetBrowserOnTimeout (no resolved browser exists to forget in that case).
bitbay
added a commit
to bitbay/chrome-devtools-mcp
that referenced
this pull request
Sep 28, 2026
…connection Re-implemented as a fresh diff against upstream's current ToolHandler.ts (post-ChromeDevTools#2659/ChromeDevTools#2787, which predates this branch's whole timeout-wrapping fix), not a literal merge of this branch's prior diff — unrelated upstream divergences (.strict() input schema, `disabled` field, `handle` as a class-field arrow function using `using _guard = ...`) are left as-is. Replaces the ToolHandler.ts portion of this branch's 6d2d4e9, and f67b230 in full: - TOOL_CALL_TIMEOUT_MS, ToolCallTimeoutError, and #raceWithTimeout<T>() ported verbatim. Puppeteer gives no way to cancel a pending call, so a timeout only stops the handler from waiting — the underlying attempt keeps running abandoned, which is exactly why the onTimeout callbacks below exist. - getContext() is wrapped in #raceWithTimeout, calling the injected abandonPendingBrowserAttemptOnTimeout on timeout (see the BrowserManager commit) so a resolution nobody is waiting for anymore doesn't get silently installed for a later caller. - The tool.handler() + response.handle() sequence is wrapped in a single shared budget (matching f67b230's reasoning: several tools' actual CDP calls happen inside response.handle(), not tool.handler(), so both need covering by one timeout or the second phase hangs unprotected after the first one's already accounted for). On timeout, forgetBrowserOnTimeout is called with the resolved context's browser. - src/index.ts: threads BrowserManager's forget()/abandonPendingAttempt() down as bound closures into ToolHandler's two new required constructor params (`forgetBrowserOnTimeout`, `abandonPendingBrowserAttemptOnTimeout`) instead of the old free-function defaults browser.ts used to provide, since BrowserManager is now private instance state owned by McpServer rather than importable module state. #getContext() itself needs no change — it already calls this.#browserManager.ensureBrowser(). This branch's 3d9dd88 (AGENTS.md style fixes: no `!`, no hand-rolled mocks, sinon.assert over Node's assert) has nothing to port forward — this diff was written directly against those rules and upstream's current tests/mocks.ts conventions, so there was nothing to retrofit. Tests: ~23 existing `new ToolHandler(...)` call sites updated for the two new required constructor params (no-op sinon.spy() stand-ins where the test doesn't care about them). Three new regression tests ported from this branch's prior versions: a hung tool.handler() times out and forgets the browser; a hung response.handle() times out even when tool.handler() resolves fast; a hung getContext() times out and abandons the pending attempt without touching forgetBrowserOnTimeout (no resolved browser exists to forget in that case).
bitbay
added a commit
to bitbay/chrome-devtools-mcp
that referenced
this pull request
Sep 29, 2026
…connection Re-implemented as a fresh diff against upstream's current ToolHandler.ts (post-ChromeDevTools#2659/ChromeDevTools#2787, which predates this branch's whole timeout-wrapping fix), not a literal merge of this branch's prior diff — unrelated upstream divergences (.strict() input schema, `disabled` field, `handle` as a class-field arrow function using `using _guard = ...`) are left as-is. Replaces the ToolHandler.ts portion of this branch's 6d2d4e9, and f67b230 in full: - TOOL_CALL_TIMEOUT_MS, ToolCallTimeoutError, and #raceWithTimeout<T>() ported verbatim. Puppeteer gives no way to cancel a pending call, so a timeout only stops the handler from waiting — the underlying attempt keeps running abandoned, which is exactly why the onTimeout callbacks below exist. - getContext() is wrapped in #raceWithTimeout, calling the injected abandonPendingBrowserAttemptOnTimeout on timeout (see the BrowserManager commit) so a resolution nobody is waiting for anymore doesn't get silently installed for a later caller. - The tool.handler() + response.handle() sequence is wrapped in a single shared budget (matching f67b230's reasoning: several tools' actual CDP calls happen inside response.handle(), not tool.handler(), so both need covering by one timeout or the second phase hangs unprotected after the first one's already accounted for). On timeout, forgetBrowserOnTimeout is called with the resolved context's browser. - src/index.ts: threads BrowserManager's forget()/abandonPendingAttempt() down as bound closures into ToolHandler's two new required constructor params (`forgetBrowserOnTimeout`, `abandonPendingBrowserAttemptOnTimeout`) instead of the old free-function defaults browser.ts used to provide, since BrowserManager is now private instance state owned by McpServer rather than importable module state. #getContext() itself needs no change — it already calls this.#browserManager.ensureBrowser(). This branch's 3d9dd88 (AGENTS.md style fixes: no `!`, no hand-rolled mocks, sinon.assert over Node's assert) has nothing to port forward — this diff was written directly against those rules and upstream's current tests/mocks.ts conventions, so there was nothing to retrofit. Tests: ~23 existing `new ToolHandler(...)` call sites updated for the two new required constructor params (no-op sinon.spy() stand-ins where the test doesn't care about them). Three new regression tests ported from this branch's prior versions: a hung tool.handler() times out and forgets the browser; a hung response.handle() times out even when tool.handler() resolves fast; a hung getContext() times out and abandons the pending attempt without touching forgetBrowserOnTimeout (no resolved browser exists to forget in that case).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Prefactoring for the SDK v2.