Rename 'Edit file' to 'Edit existing file' in docs

[luislobo9b]

Summary by cubic

Updated Agent mode docs to use the correct tool identifier, changing edit_file to edit_existing_file for the “Edit file” tool so users call the right command.

^{Written for commit 722127b. Summary will update automatically on new commits.}

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[cubic-dev-ai]

No issues found across 1 file

[RomneyDa]

@luislobo9b looks good! Note CLA comment!

[RomneyDa]

@luislobo9b note you will likely need to rebase for this to pass tests, there were some issues on main

[BekahHW]

@luislobo9b check out #8777 (comment). It has the approval, once you address the issues Dallin mentioned.

[cubic-dev-ai]

15 issues found across 136 files (reviewed changes from recent commits).

Prompt for AI agents (all 15 issues)

Understand the root cause of the following 15 issues and fix them.


<file name="gui/src/pages/gui/Chat.tsx">

<violation number="1" location="gui/src/pages/gui/Chat.tsx:280">
The feedback dialog check runs against the pre-increment value, so it doesn’t fire until the 301st input instead of the intended 300th. Compare against the incremented counter value before triggering the dialog.</violation>
</file>

<file name="core/autocomplete/util/AutocompleteLruCache.ts">

<violation number="1" location="core/autocomplete/util/AutocompleteLruCache.ts:59">
The singleton promise is never reset when initialization fails, so a single startup error permanently breaks AutocompleteLruCache.get() for the rest of the process.</violation>

<violation number="2" location="core/autocomplete/util/AutocompleteLruCache.ts:180">
Dirty entries are cleared before confirming a successful flush, so any failed transaction permanently drops those changes instead of retrying them.</violation>
</file>

<file name="docs/ide-extensions/autocomplete/how-it-works.mdx">

<violation number="1" location="docs/ide-extensions/autocomplete/how-it-works.mdx:27">
`/customize/models#autocomplete` is a broken target for the “Autocomplete deep dive” link—no such anchor exists on that page, so the link no longer reaches the deep-dive content. Please point to the actual Autocomplete deep dive document.</violation>
</file>

<file name="extensions/cli/src/services/MCPService.ts">

<violation number="1" location="extensions/cli/src/services/MCPService.ts:182">
The token refresh wrapper caches the new OAuth token but never rebuilds or reconfigures the active MCP transport, so the retried operation still sends the stale Authorization header and continues to fail with 401. Reconnect the client or update the transport headers after refresh before retrying.</violation>
</file>

<file name="gui/src/pages/AddNewModel/configs/models.ts">

<violation number="1" location="gui/src/pages/AddNewModel/configs/models.ts:1070">
The three new Cohere Command A model entries you added are never registered with the `cohere` provider, so they remain unreachable in the UI. Please add them to `providers.ts`&#39;s `cohere.packages` list so users can actually select them.</violation>
</file>

<file name=".github/workflows/pr-checks.yaml">

<violation number="1" location=".github/workflows/pr-checks.yaml:338">
Pin the `jlumbroso/free-disk-space` action to a specific commit SHA instead of the mutable `@main` reference to avoid supply-chain risk from unreviewed upstream changes.</violation>
</file>

<file name="extensions/cli/src/services/mcpUtils.ts">

<violation number="1" location="extensions/cli/src/services/mcpUtils.ts:12">
Classifying HTTP 405 responses as authentication errors forces an unnecessary token refresh and prevents the transport-mismatch fallback the MCP client expects. Remove the 405 checks so only real auth failures (401/Unauthorized) trigger a refresh.

(Based on your team&#39;s feedback about allowing SSE fallback when the initial connect returns 405.) [FEEDBACK_USED]</violation>
</file>

<file name="extensions/cli/src/services/AgentFileService.ts">

<violation number="1" location="extensions/cli/src/services/AgentFileService.ts:61">
Swallowing the hub fetch exception here hides the real reason an agent slug failed to load—every network/auth/404 error is transformed into a misleading local file error (usually ENOENT), so users cannot diagnose real hub problems.</violation>
</file>

<file name="core/config/profile/doLoadConfig.ts">

<violation number="1" location="core/config/profile/doLoadConfig.ts:355">
Restricting the telemetry opt-out logic to VS Code ignores other IDEs’ telemetry settings, so their host-level opt-out is no longer enforced. Remove the ideType check so every IDE’s isTelemetryEnabled() result can disable anonymous telemetry when the host preference is off.</violation>
</file>

<file name="core/autocomplete/util/AutocompleteLruCache.test.ts">

<violation number="1" location="core/autocomplete/util/AutocompleteLruCache.test.ts:642">
`sqlite.open` is mocked after AutocompleteLruCache is imported, so the singleton test still uses the auto-mocked open that returns undefined and `AutocompleteLruCache.get()` blows up before it can return an instance.</violation>
</file>

<file name="docs/package.json">

<violation number="1" location="docs/package.json:20">
`axios` is added as a docs dependency but no docs source imports it, so it’s currently unused extra surface area and install weight. Either wire it up or drop the dependency.</violation>

<violation number="2" location="docs/package.json:21">
`tar-fs` is being pulled into the docs package without any code using it, adding unnecessary dependency risk and install time. Remove it unless there’s a concrete usage.</violation>
</file>

<file name="packages/llm-info/src/providers/cohere.ts">

<violation number="1" location="packages/llm-info/src/providers/cohere.ts:9">
maxCompletionTokens must not exceed the model’s context window; requesting 8192 completion tokens on a model with an 8000-token context will fail.</violation>
</file>

<file name="core/autocomplete/CompletionProvider.ts">

<violation number="1" location="core/autocomplete/CompletionProvider.ts:315">
`dispose()` closes the shared autocomplete cache but never reinitializes it, so after one dispose every future CompletionProvider reuses a closed cache (DB closed, flush timer stopped) and caching breaks for the rest of the process.</violation>
</file>

_{Reply to cubic to teach it or ask questions. Re-run a review with @cubic-dev-ai review this PR}

[luislobo9b]

I have read the CLA Document and I hereby sign the CLA.

@RomneyDa @BekahHW

I’m learning, but I think I did it correctly

[RomneyDa]

@luislobo9b thanks for the contribution! Note the `conventional commits" PR title/commits error
https://www.conventionalcommits.org/en/v1.0.0/