A new system prompt injection that instructs Claude to investigate before asking clarifying questions, gated behind the tengu_slate_harrier feature flag and configurable via CLAUDE_CODE_INVESTIGATE_FIRST environment variable.
Feature-flagged via tengu_slate_harrier (defaults to "off"). Only applies when using claude-opus-4-7.
- Two modes:
"additive"(adds investigation guidance to the existing "executing actions with care" prompt) and"compact"(uses a shorter version of the actions prompt plus investigation guidance) - In both modes, adds: "Asking the user a clarifying question has a cost: it interrupts them... Before asking, spend up to a minute on read-only investigation (grep the codebase, check docs, search memory) so your question is specific."
- Can be overridden via
CLAUDE_CODE_INVESTIGATE_FIRST=additive|compact|off
Feature flag check (search for "tengu_slate_harrier") and env var (search for "CLAUDE_CODE_INVESTIGATE_FIRST")
Strings lifted out of the shipped bundle, so the claim above can be checked against them.
tengu_slate_harrier Not enough to sayNothing 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 v2.1.122: no gate table built for this version
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 v2.1.122. It isn't a statement about your account. What a flag value here can and cannot tell you
A reading is one sample. Claude Code evaluates its flags remotely, so no client sees the targeting rule behind a value and this says nothing about your account.
A reading is one sample. Claude Code evaluates its flags remotely, so no client sees the targeting rule behind a value and this says nothing about your account.