code
A Claude Code settings key, read out of the shipped bundle. It was in the build from v2.1.301 and was gone by v2.1.302.
MCP auth and HTTP failures are classified more accurately, and cancelled tool calls no longer trigger slow-call warnings.
Not Anthropic's. This is the line from the earliest changelog entry here that named it, v2.1.221.
In the changelogs
32Releases whose published page names code.
-
v2.1.296
Gateway warns when settings will not reach Claude Desktop's Code tab
The Claude apps gateway now warns that only settings inside a
code:block reach Claude Desktop's Code tab, even withserve_to_desktop: truefound in this entry's text
-
v2.1.296
Gateway policy files gain a
codeblock for Claude Desktop's Code tabGateway policy entries can now use a code block for Claude Desktop's Code tab, but it cannot be mixed with cli or settings blocks
found in this entry's text
-
v2.1.296
Managed policy entries can carry a
codesection as well ascliEntries in
managed.policiesnow accept acodekey in addition tocli, and validation messages point to whichever one the entry usesfound in this entry's text
-
v2.1.296
Desktop gateway warning now tells admins exactly what to change
The Desktop gateway warning now says to set
serve_to_desktop: trueon the user's policy, or changeclitocodein every policyfound in this entry's text
-
v2.1.290
More Anthropic sites count as trusted documentation hosts
Claude Code's list of trusted documentation hosts now includes
claude.devand Anthropic's experimental and raw GitHub locationsfound in this entry's text
-
v2.1.285
Claude Code help skill can only fetch pages from Claude's docs sites without asking
The built-in skill that answers questions about Claude Code now has web fetching approved only for code.claude.com and platform.claude.com
found in this entry's text
-
v2.1.281
/model error messages get codes and wording for hosts without the command
Failed model switches now return an error code and no longer tell you to run
/modelwhere it isn't availablefound in this entry's text
-
v2.1.281
Model picked at launch skips the server check
A model chosen at launch is no longer checked with the server, and rejected model switches now carry a code such as blocked_by_hook
found in this entry's text
-
v2.1.281
Model switch rejections now return a machine-readable code
A blocked model switch now returns a
codefield naming the reason, such asrestricted_by_orgorcatalog_unknownfound in this entry's text
-
v2.1.281
Setting auto permission mode now returns a machine-readable code
When switching to auto permission mode fails, the result now includes a machine-readable
codegiving the reasonfound in this entry's text
-
v2.1.281
New refusal: path spelling that session permission rules cannot match
Claude Code can now refuse a file path written in a form its permission rules cannot match and ask for the full path
found in this entry's text
-
v2.1.281
SDK model switch at launch skips server confirmation; switch errors now carry a code
A model chosen through the SDK at launch is no longer checked with the server first, and model-switch errors now include a code
found in this entry's text
-
v2.1.280
Structured error objects with a code+description now get a dedicated short summary
Error messages with a code and description now show a short, dedicated summary instead of a scanned-out file path or command
found in this entry's text
-
v2.1.280
Working-directory registration failures now carry a machine-readable error code
Working-directory registration failures now include a machine-readable error code alongside the human-readable reason
found in this entry's text
-
v2.1.280
Denied path reason now carries a machine-readable code
Denied-path errors now include a machine-readable 'outside_scope' code
found in this entry's text
-
v2.1.273
AWS error parsing now also reads details.error_code
AWS server error parsing now also reads error codes nested under
details.error_codefound in this entry's text
-
v2.1.261
Plugin command failures now tagged with claude.ai error codes
Plugin command failure telemetry now tags claude.ai-specific error codes.
found in this entry's text
-
v2.1.246
Artifact request failures say when the sandbox proxy blocked them
A blocked network request no longer looks like an artifact permissions problem.
found in this entry's text
-
v2.1.235
Account-on-hold handling is built end-to-end but dark behind
tengu_lively_beaver(fallback: off)A full account-on-hold sign-in flow was added but stays unreachable pending a server-side gate
found in this entry's text
-
v2.1.235
MCP tool permission dialogs render parameters as a key/value table
MCP tool permission dialogs now render parameters as a formatted key/value table with links and timestamps
found in this entry's text
-
v2.1.232
Plugin names are sanitized before appearing in failure messages
A hostile plugin name can no longer inject escape codes into failure messages in your terminal.
found in this entry's text
-
v2.1.232
Code blocks in HTML artifacts get syntax highlighting
Code blocks in generated HTML artifacts now get syntax highlighting, within a size budget.
found in this entry's text
-
v2.1.221
MCP error classification recognises typed HTTP and OAuth errors
MCP auth and HTTP failures are classified more accurately, and cancelled tool calls no longer trigger slow-call warnings.
found in this entry's text
-
v2.1.221
MCP HTTP auth errors matched on both code and errorCode
Claude recognizes MCP login failures more reliably, so auth errors are handled instead of retried blindly.
found in this entry's text
-
v2.1.221
MCP OAuth refresh retries on error codes and preserves issuer and redirect URI
MCP OAuth token refresh retries smarter and keeps your saved redirect URI across reconnects.
found in this entry's text
-
v2.1.203
claude.com/docs Added to WebFetch Pre-Approved Domains
found in this entry's text
- v2.1.110
-
v2.1.84
"Open in Claude Code on the Web" Action
found in this entry's text
- v2.1.19
-
v2.1.0
OAuth URL Parameter Redaction
found in this entry's text
-
v1.0.114
Smart Editor Detection
found in this entry's text
-
v0.2.106
Enhanced macOS Terminal Detection
found in this entry's text
First cited
0No release's published evidence quotes this name. The ledger indexes the Evidence lines of every changelog on this site, so this says nobody here has ever quoted it; it says nothing about how old the name is.
Presence across releases
2Build by build
8One row per release since the first build this name was read out of. Absent means the build was read and the name was not in it, never mined means no bundle for that release was ever archived, and a declared type or a default is only ever what that release's own bundle stated.
Read out of the published npm bundle release by release, and out of Anthropic's own documentation as this site captured it. Nothing on this page is a description anybody here wrote about what the settings key does. All settings keys.