Source Intelligence

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 of v2.1.84 Home All releases olderv2.1.83 v2.1.85newer
Claude Code v2.1.84

Agent List in System Messages [Gradual Rollout]

Feature flag
tengu_agent_list_attach Off in both readings

The flag server returned off for the account this site reads and for the anonymous baseline. A reading of off cannot rule out a rollout these two readings sit outside of.

This account: off · anonymous baseline: off · compiled default in v2.1.84: 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 v2.1.84. It isn't a statement about your account. What a flag value here can and cannot tell you

What

Agent type descriptions can be injected directly into system-reminder messages instead of only in the Task tool description.

Status

Feature-flagged via tengu_agent_list_attach (default: false). Can be overridden with CLAUDE_CODE_AGENT_LIST_IN_MESSAGES environment variable.

Details
  • When enabled, available agent types are listed in <system-reminder> messages in the conversation
  • Agent type descriptions use the format: - agentType: whenToUse (Tools: toolList)
  • This helps models discover available agent types more reliably
Evidence

Agent list feature (gated by tengu_agent_list_attach, search for "CLAUDE_CODE_AGENT_LIST_IN_MESSAGES")

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

See this entry in the whole of v2.1.84 →