Source Intelligence
Follow Discord

DisclaimerUnofficial, and not affiliated with Anthropic. Nearly all of this is read straight out of what ships: npm bundles, captured prompts, published docs. Anthropic's own notes go in verbatim, marked as theirs. The rest is my reading, and every entry carries the strings behind it. If one looks wrong, vote it down and say why.

← All releases Home ←olderv2.1.176 v2.1.178newer→

Claude Code v2.1.177

3 entries read diff v2.1.176 → v2.1.177 Markdown

This is a small patch release focused on internal infrastructure. The main addition is a new model error override system that allows server-side customization of error messages and model picker hints. The release also includes a fix that restricts a first-party-only update pathway to first-party installations.

Find
Pick an entry · j / k steps through
3 entries

Improvementsopen

Model Error Override System#

Infrastructure was added that allows server-side configuration to customize the error messages and model-picker hint text that users see when a model encounters an error.

Details
  • A new config key tengu-model-error-overrides is read from server-delivered configuration. It maps model identifiers to objects with optional block (error message text) and pickerHint (hint shown in the model picker) fields.
  • When a matching override is present and its block string is non-empty, the override text replaces the default error display.
  • pickerHint can provide a separate, shorter hint for use in the model selection UI. If absent, the block text is reused.
  • If any error occurs reading the override config, the system logs model-error-overrides picker hint failed: or tengu-model-error-overrides block check failed: and falls back to default behavior.

This is a server-controlled capability with no user-facing knob; the overrides are delivered via the existing feature-flag config mechanism and take effect automatically when present.

Evidence

New config key (search for "tengu-model-error-overrides")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

First-Party Guard on Saffron Lattice Eligibility#

Feature flag
tengu_saffron_lattice Not enough to say

Nothing here resolved what this flag was doing on this version, so nothing here should be read as on or off.

This account: no value returned · anonymous baseline: no value returned · compiled default in v2.1.177: not a boolean we can read

These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.

Read once, for one account on one subscription tier, against v2.1.177. It isn't a statement about your account. What a flag value here can and cannot tell you

The function that checks whether a user is eligible for the saffron lattice (progressive-rollout update) feature now requires the installation to be a first-party build (l6() !== "firstParty" returns early). Previously this check was absent, meaning third-party or forked builds could inadvertently enter the saffron lattice update path.

Evidence

New guard in eligibility check (search for "firstParty" near tengu_saffron_lattice)

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Saffron Lattice Default State#

Feature flag
tengu_saffron_lattice Not enough to say

Nothing here resolved what this flag was doing on this version, so nothing here should be read as on or off.

This account: no value returned · anonymous baseline: no value returned · compiled default in v2.1.177: not a boolean we can read

These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.

Read once, for one account on one subscription tier, against v2.1.177. It isn't a statement about your account. What a flag value here can and cannot tell you

The default value used when the tengu_saffron_lattice feature flag is absent or fails to parse was changed from an empty object {} to an explicit disabled state { enabled: false }. This makes the fallback behavior unambiguous — if the server config is unavailable, the feature is definitively off rather than potentially entering an indeterminate state.

Evidence

Default sentinel zJ4 = { enabled: !1 } (search for "tengu_saffron_lattice")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Verbatim
No official entry

Not individually listed upstream

v2.1.177 does not have its own entry in Anthropic’s official changelog, because prerelease builds are often folded into a neighbouring release. Browse the full changelog for the closest official notes. Everything else on this page came out of the bundle instead, which is why the two lists don't match.

Read it on GitHub →
System prompt

No change to the system prompt since v2.1.176.

Claude Code, interactive mode

Read the prompts for this release →