@cloudflare/think@0.18.0

Minor Changes

• #2196 ec93caf Thanks @mattzcarey! - Replatform Think conversation storage onto agents/sessions and prompt context onto agents/context.

  Existing subclasses keep compiling and running. configureSession(session) still accepts the withContext() / withCachedPrompt() chain, this.session still carries addContext, getContextBlock, replaceContextBlock, refreshSystemPrompt, freezeSystemPrompt, tools(), and the rest, and appendMessage(message, parentId) / getHistory(leafId) still take their positional arguments. Those context methods are deprecated forwards to the new this.context; new code should declare blocks in configureContext() and read this.context directly. withCachedPrompt() is a no-op because the frozen prompt is always persisted now.

  What you may want to change

  • Declare prompt blocks with the new configureContext() hook instead of withContext(). Blocks from both are merged, configureContext() first.
  • Read context through this.context (a ContextBlocks from agents/context) instead of this.session.
  • Import Session from @cloudflare/think when you annotate a configureSession override. The class exported from agents/sessions is the raw storage handle and is not assignable to Think's.

  Behaviour changes on upgrade

  • Storage migrates on first wake and cannot be rolled back. Each Durable Object lifts its assistant_messages, assistant_compactions, and assistant_config rows into the cf_agents_session_* tables, verifies every row landed, and drops the old tables. An object that has woken on this version has an empty conversation if you roll back to the previous release; rolling forward again is safe. Deploy behind a canary if you need a rollback path.
  • hydrationByteBudget defaults to 32 MiB (was 24 MiB) and is now a hard ceiling that charges each row its full stored size, attachments included. There is no message-count floor, so an unusually large recent window can hydrate fewer than four messages. getHistory() still reads the full path.
  • Context blocks load during onStart, as before the replatform, so this.context.getBlock() answers as soon as the object has started.
  • session.search() still works. Its index is now built by the first search on an object rather than maintained on every append, so a Think that never searches stops paying a second billed row per message.
  • Think no longer reads Sessions tables with raw SQL. If you queried assistant_messages yourself, use this.session.history() or getHistory().

  Removed

  • sessionAttachments. There is nothing to configure about how Sessions stores a message: media leaves the row into a content-addressed attachment store and a message larger than one SQLite row is split across continuation rows, losslessly.
  • getRecentHistory(budget, minRecentMessages): the second argument is accepted and ignored.
  • compactAfter(threshold, { tokenCounter }): the counter option is accepted and ignored; the trigger reads the estimate Sessions stamps on each row.

• #2175 8ffb3ad Thanks @mattzcarey! - Lifecycle owns a durable job queue, driven as an alarm event loop.

  The thing in the queue is a job: a serialisable callback address — the
  owning capability plus a function name — with a due time and a payload.
  Capabilities and the host push jobs through the scoped jobs surface;
  Lifecycle drives due jobs in timestamp order when the alarm fires, owns
  dispatch retries and platform-failure deferral, arms a deadman pre-alarm
  before driving so an isolate death mid-drive still wakes the object, and
  derives the physical alarm purely from queue state (queue mutations re-arm
  automatically; an exclusive job suppresses ordinary candidates).

  class Cleanup extends LifecycleCapability {
    async scheduleSweep(time: number) {
      await this.lifecycle.jobs.push({ id: "sweep", fn: "sweep", time });
    }
    onJob({ job }: LifecycleJobContext) {
      // drive result: nothing = complete, { rescheduleAt } = suspend,
      // "yield" = leave due and wake again immediately
    }
  }

  The pull-based alarm-contribution model is removed: capability
  getNextAlarm()/onAlarm(), host getNextAlarm(),
  LifecycleServices.alarms (rearm/disabled), and AlarmContribution
  are gone. Host onAlarm() remains and runs once per alarm invocation
  after due jobs are driven. Terminal application failures reach the
  owner's onJobError(), whose drive result decides advancement.

  The alarm memory-limit circuit breaker (#1825) moves from Agent.alarm()
  into the Lifecycle event loop, targeting the exact executing job; Agent
  contributes domain policy through the new onAlarmMemoryLimit() host
  hook, and Scheduler's __DO_NOT_USE_WILL_BREAK__handleAlarmMemoryLimit
  escape hatch is gone. After recording a strike the breaker now finishes by
  resetting the isolate with ctx.abort(reason, { retryAlarm: false })
  (retry of the handled alarm suppressed; the backoff alarm owns the next
  wake), and Agent.destroy() uses the same no-retry abort so a completed
  teardown's alarm cannot be retried into a fresh constructor that recreates
  the deleted schema.

  Scheduler keeps its entire public API and loses its storage and due-row
  loop: a schedule is one job whose fn is the callback name, and interval
  schedules are single-flight jobs. Existing cf_agents_schedules rows are
  migrated into the cf_agents_jobs queue on startup and the legacy table
  is dropped. Agent's public scheduling and keepAlive() APIs are
  unchanged; its keep-alive, fiber-recovery/facet housekeeping, and
  deferred-destroy wakes are now host jobs, and Think's
  workflow-notification wake replaces the removed _getExtensionAlarm().

• #2196 ec93caf Thanks @mattzcarey! - Keep media eviction a context-window technique, separate from how Sessions stores a message.

  How Sessions lays a message out in rows is invisible and lossless: a message too large for one row is split across continuation rows and reassembled byte for byte. Media eviction is a decision about the model's context: once media has aged past mediaEviction.keepRecentMessages on the active path, Think removes it from the conversation so the model stops re-reading a large image every turn, and leaves [evicted image/png, 812004 bytes; preserved at /attachments/evicted/<messageId>-<n>.png] in its place. The raw bytes are written to the Workspace at that path with their real mime type, so the workspace read tool puts the actual image back in context when the agent deliberately reads it. The rewritten message no longer carries the payload, so the bytes live in exactly one place.

  mediaEviction: false now means the model keeps seeing aged media. It no longer changes where Sessions keeps the bytes.

  minPartBytes is Think's context threshold and is no longer passed to Sessions as a storage setting. The marker text and the /attachments/evicted/<id>-<n>.<ext> paths are unchanged, so old markers keep resolving.

  MediaEvictionConfig.externalizeToWorkspace is deprecated and ignored: evicted bytes are always preserved. Existing configurations keep compiling.

  WorkspaceLike.writeFileBytes is optional. Eviction and skills projection need it to write raw bytes; a custom workspace without it keeps working and those two features log once and stand down.

• #2216 dd09d44 Thanks @mattzcarey! - feat(streams): rollover block log and an atomic stream → message cutover; no more stream-buffer sweeps.

  The Streams chunk log is now mutable rollover blocks: an append grows the open block row (an UPDATE) until it reaches 256 KB, then opens the next. Same one billed row per append as before, but a stream of thousands of chunks is a handful of rows to delete instead of thousands. Existing cf_agents_stream_chunks rows are folded into blocks lazily, one stream at a time on first touch, so startup never reads the whole legacy log; the table is dropped once it is empty.

  writer.close({ commit, discard }) (and error(reason, { … })) settles the stream, runs the caller's synchronous writes and deletes the stream's rows in one SQLite transaction. Session.__DO_NOT_USE_WILL_BREAK__sync().upsert() is the matching synchronous message write; its after() dispatches the change feed and auto-compaction once the transaction commits.

  Chat hosts (AIChatAgent, Think) now persist the finished turn's assistant message inside that cutover: the message, the stream's settlement and the deletion of its temporary rows commit together, so a crash leaves either the live stream (recovery rebuilds the message from it) or the message, never neither. ResumableStream.start() reclaims anything a crash left behind. The _cleanupStreamBuffers alarm is no longer armed (cleanupStreamBuffers and STREAM_CLEANUP_DELAY_SECONDS are removed from agents/chat; the host callback is kept as a no-op so alarms persisted by earlier versions still resolve).

Patch Changes

• #2194 6da4c44 Thanks @mattzcarey! - Run root-agent chat recovery continuations as chained Tasks instead of schedule rows. Initial recovery attempts deduplicate by incident, delayed retries use durable Task sleeps, and platform failures replay through Task claims. AI Chat and Think share one reserved recovery definition and preserve their existing bounded callback handoff behavior: a failure before handoff stays with the current queue execution, while a detached post-handoff platform failure enqueues exactly one replacement.

  Tasks now propagate condemned-isolate failures out of journaled steps and apply alarm memory-limit backoff and sealing to the run whose wake struck — claim stripped and deadline pushed, so startup reconciliation cannot resurrect it and the reclaim still sees an interrupted attempt. Task wake jobs are pushed with a single dispatch attempt so a platform failure rejects the alarm instead of being retried into a silent reschedule of the still-claimed run. Lifecycle gains trackAlarmWork(): work a job hands off at a bounded return stays inside that alarm's memory-limit breaker domain after the alarm returns, so other jobs stay live while a memory reset from the handoff still records a strike — one strike per reset however many flows observe it — and strikes clear only once no handed-off work is outstanding and the last of it settled clean. retain: false now removes failed and cancelled runs as well as completed runs, releasing journals and idempotency keys after every terminal outcome. Routed dynamic agents temporarily retain the root-owned schedule transport until Tasks supports routed child wakes.

  AI Chat and Think require agents >=0.23.0, the pending release batch containing the shared recovery Task definition and internal enqueue support.

• #2173 71ce28a Thanks @mattzcarey! - Replatform chat's resumable streams onto the agents/streams capability.

  ResumableStream is now a thin adapter over Streams: chat's in-flight turn output lives in the shared durable chunk log (cf_agents_streams / cf_agents_stream_chunks), packed ~10 wire chunks per stored segment for write economy, with completion/error mapped onto stream settlement and retention keyed off the stream row's updated_at (sweeps no longer scan the chunk table). Existing cf_ai_chat_stream_* tables migrate wholesale — including an in-flight stream — on first construction after upgrade, then are dropped. AIChatAgent and Think expose the backing capability as readonly streams, so any streams.read() consumer on the same Durable Object can observe chat streams. The chat wire protocol, replay handshake, and recovery behavior are unchanged.

• #2223 dd8bf90 Thanks @mattzcarey! - perf(chat): derive the recovery forward-progress marker from the stream log instead of bumping a KV counter per credited chunk. ResumableStream.progressMarker() counts durably flushed segments — live streams from their log tails, deleted streams from a retired total folded in as their rows are removed — so the marker stays monotonic across cutover and reclaim, never moves on a reconnect replay or a recovery re-persist, and ignores compaction. A parent forwarding a sub-agent's output credits it explicitly through creditProgress(). Nothing is written per chunk any more; one row is written per stream retired. The old KV counter is read once per isolate and seeded into the marker so an in-flight incident never sees it drop, and the hosts mirror the marker's durable part back to that key per stream retired, so a rollback reads no lower either. The Streams sync aperture gains an onDelete hook so a chat row deleted through the public capability is retired like any other. AIChatAgent now flushes a settled tool result to SQLite the moment it is stored, as Think already did, so it is durable before the next packed flush and counts as progress immediately. The work budget's unit is now the durable segment, and DEFAULT_CHAT_RECOVERY_MAX_WORK moves from 1000 to 10000 to stay as generous as before for delta-heavy turns. Two cutover fixes ride along: a Think agent-tool child now keeps its stream rows for the parent to tail after completion, as ai-chat already did, and the Streams capability re-derives its legacy-table flag after a rolled-back cutover.

• #2219 0966a0b Thanks @mattzcarey! - perf(think): stop re-reading the transcript during a turn. A tool update (client result, approval, cross-message result, execution outcome) used to read the whole persisted history to find its one target message; it now resolves the owner from the in-flight accumulator and the live cache and reads that row alone, walking storage newest-first only when the cache does not cover the active path. A chat request no longer reads the path twice and upserts every echoed message: the cache is the server transcript, unchanged messages are skipped before Sessions sees them, and only what changed is written. The cache re-windows itself once appends carry it past hydrationByteBudget, and marks itself stale on Sessions import and compaction events. Media eviction is scheduled from an in-memory check on each linear append instead of a stored-path scan after every cache refresh, think_config rewrites of an unchanged request body or client-tool schemas are skipped, and the agent-tool child-run DDL runs once per isolate.

• #2190 58c586a Thanks @mattzcarey! - Make the alarm memory-limit circuit breaker (#1825) a self-contained
  Lifecycle concern instead of an Agent-mediated one.

  Recovery-loop membership is now a property of the job row
  (LifecycleJobPushOptions.recoveryLoop): flagged jobs are backed off by
  the breaker on a strike and purged when it seals at the strike budget,
  without disturbing unrelated rows — a recovery schedule can no longer
  silently escape the breaker. The public ScheduleOptions vocabulary is
  unchanged: schedules only shape future work, and chat recovery reaches the
  flag through internal scaffolding (RecoveryLoopScheduleOptions) retained
  only for legacy rows and routed dynamic agents. Root recovery moves to Tasks;
  the scaffolding can be deleted when Tasks supports routed child wakes.
  Capabilities can react to a strike through the new optional onMemoryLimit
  hook, hosts through onAlarmMemoryLimit, and the context identifies the job
  that was executing when one exists. The strike budget is real Lifecycle
  configuration (Lifecycle.install(host, { maxAlarmMemoryLimitStrikes }))
  rather than a composition-root side channel. Until Tasks supports routed
  child wakes, a sealed recovery schedule also forwards the seal to its owning
  dynamic agent so a chat child under a plain Agent root persists its exhausted
  incident and terminal notification.

  Removed accordingly: Agent.onAlarmMemoryLimit's policy relay, the
  _cf_recoveryAlarmCallbacks template hook, Scheduler.applyMemoryLimitPolicy,
  and
  setLifecycleAlarmMemoryLimitStrikes. AIChatAgent and Think flag their
  routed recovery fallback via chatRecoverySchedulePolicy and seal in-flight
  incidents from their own protected onAlarmMemoryLimit hooks; both now
  require agents >= 0.23.0 from the pending release batch (they consume its
  new agents/chat recovery exports and no longer implement the old
  template-method breaker hooks). Agent retains a
  sealed-only call to _cf_sealMemoryLimitedRecovery so already-published chat
  packages whose peer ranges accept agents 0.23 keep terminal notifications;
  that fallback carries no callback-name or queue policy.

• #2230 e18d42f Thanks @ben-reitz! - Pass proactively compacted messages to beforeStep so hooks that append context do not restore the original history. Explicit message overrides still take precedence.