Follow Discord
Sweep 22 Oct 2026 · 21:04Z Build v2.1.309 528 read Stable v2.1.308 Latest v2.1.308 Next v2.1.309 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.101 ·

/loop Dynamic Self-Pacing Mode [In Development]

Entry
JSON All of v2.1.101
EntryKind: in v2.1.101,
In DevelopmentSection of the release
What

A major enhancement to the /loop command that allows the model to dynamically choose its own iteration timing instead of requiring a fixed interval.

Status

Feature-flagged via tengu_kairos_loop_dynamic (defaults to false) and tengu_kairos_loop_prompt (defaults to false)

Details
  • When enabled, /loop can be invoked without an interval, and the model decides how long to wait between iterations based on the task context
  • Uses a new ScheduleWakeup tool for the model to request its next wake-up with a specific delay (clamped to 60–3600 seconds)
  • Includes sophisticated cache-aware delay selection: delays under 270s keep the prompt cache warm, delays over 300s incur a cache miss
  • Supports loop.md task files in .claude/loop.md or ~/loop.md that define recurring task instructions
  • Introduces sentinel values <<autonomous-loop-dynamic>> and <<loop.md-dynamic>> for the dynamic pacing variants
  • Includes a detailed autonomous loop system prompt that instructs the model on how to behave during timer-based autonomous operation, including PR maintenance, CI monitoring, and scope management
  • The system prompt emphasizes maintaining user trust: "Acting on what the conversation already established is safe and valuable. Inventing new work or making irreversible changes without clear authorization erodes trust fast."
  • Loops have a maximum age limit, after which they automatically age out
Evidence

Dynamic loop system (gated by tengu_kairos_loop_dynamic and tengu_kairos_loop_prompt, search for "Autonomous-loop instructions")

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

Read from
Feature flag
tengu_kairos_loop_dynamic On for this account, and not off by default

The flag server returned on for the one account this site reads, and nothing in this release compiles it off by default. The compiled default is shown below, and says which it is when we cannot read one: a fifth of gates compile in a string or a number rather than on or off, and most published releases have no gate table behind them at all. No client can see what the server returns for your account.

This account: on · anonymous baseline: on · compiled default in v2.1.101: 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.

tengu_kairos_loop_prompt On for this account, and not off by default

The flag server returned on for the one account this site reads, and nothing in this release compiles it off by default. The compiled default is shown below, and says which it is when we cannot read one: a fifth of gates compile in a string or a number rather than on or off, and most published releases have no gate table behind them at all. No client can see what the server returns for your account.

This account: on · anonymous baseline: on · compiled default in v2.1.101: 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 v2.1.101. It isn't a statement about your account. What a flag value here can and cannot tell you

What has happened since
Flag reading moved The flag server now returns on for tengu_kairos_loop_dynamic, read as this 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.
Flag reading moved The flag server now returns nothing at all, having stopped carrying it for tengu_kairos_loop_dynamic, read as this 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.

See this across every release →

See this entry in the whole of v2.1.101 →

Feedback