@flue/runtime
2.3.1 · 18 Nov 2026Apache-2.0524.4k / weekRepositoryHomepagenpmAtom feed
Upgrading @flue/runtime From 0.11.0 to 2.3.1
17 releases in betweenevery note read
- 33
- changes to make
- 0
- advisories fixed
- 28
- new
- 96
- fixes
Start here
Before you upgrade
Tick each one off as you go- Breaking2.0.0
- Breaking2.0.0
- Breaking2.0.0
- Breaking2.0.0
- Breaking2.0.0
- Breaking2.0.0
- Breaking2.0.0
- Breaking2.0.0
- Breaking2.0.0
- Breaking2.0.0
Release by release
- Fixed
Updated
@earendil-works/pi-aiand@earendil-works/pi-agent-coreto 1.5.2, fixing builds that picked up pi-ai 1.5.@flue/runtime2.3.0 imported TypeBox through@earendil-works/pi-ai, which stopped re-exporting it in 1.5.0, so a fresh install failedvite build. Flue now depends ontypeboxdirectly and requires pi^1.5.0; remove any overrides that held pi at 1.4. If your own code importsTypeorStaticfrom@earendil-works/pi-ai, import them fromtypeboxinstead. Pi now validates tool arguments with its own JSON Schema validator, which coerces a little more (for example"3"for a number) and words its validation errors differently. OpenAI-compatible requests with no tools no longer send"tools":; setcompat.requiresToolsWithToolHistory: trueon models behind proxies that require it, such as older LiteLLM versions in front of Anthropic. - Fixed
Workers AI binding models on the Responses API declare tools the way pi's own OpenAI provider does. Since pi 1.4, the binding passed a boolean where pi expected a tool-support mode, so a model without deferred tool loading could leave tools added mid-conversation out of its request. Models that support mid-conversation tool additions now declare every tool through
additional_toolsinput items.
- Fixed
- New
Updated
@earendil-works/pi-aiand@earendil-works/pi-agent-coreto 1.1.0. New models resolve with no configuration, including Claude Sonnet 5.5 (anthropic/claude-sonnet-5-5), Claude Haiku 5.5 (anthropic/claude-haiku-5-5), and GPT-6.1 Sol (openai/gpt-6.1-sol). Azure users must update: Pi renamed the Azure provider fromazure-openai-responsestoazure, so changeazure-openai-responses/...model specifiers andprovidersconfig entries toazure. TheAZURE_OPENAI_*environment variables are unchanged. Tool and turndurationMsvalues now come from Pi's monotonic timing, so a tool's duration covers only its execution, not Flue's hooks or result storage. - Fixed
useAgentFinishcallbacks once again see state written during the response. Since 2.2.0, finish hooks ran against the agent render from before the response's final turn, so a durable guard set by a finish callback — the documentedusePersistentStatepattern for appending a reminder once — was never observed: the callback re-appended on every cycle until the response failed at the runaway-loop ceiling. The agent now re-renders before each finish-hook cycle (#818). - Fixed
read_skill_resourceaccepts a packaged skill path without its leading/. Models often echo the advertised/.flue/packaged-skills/...path without the slash, and the read failed as not found (#781). - Fixed
The local sandbox SIGKILLs a command's whole process group after the grace period, even when the shell exits first. On an
exec()timeout or abort, the pending SIGKILL was cancelled once the shell's pipes closed, so a process that ignored SIGTERM and held none of the shell's output kept running afterexec()returned (#809). - Fixed
Execution interceptors see
agentNameon model, tool, and shell steps, not only on the outeragentoperation. The root harness's execution context carried only the instance id, so an interceptor that branches per agent took its default branch on every step (#785). - Fixed
flue init --target cloudflarescaffolds the Workers AI binding that keylesscloudflare/...models need, commented out inwrangler.jsonc. The generated agent suggested switching to a keyless Cloudflare model but nothing declaredenv.AI; it now points at the binding to uncomment. The binding stays off by default because it reaches your Cloudflare account even in local dev (#739). - Fixed
The E2B sandbox blueprint returns a non-zero exit from
exec()instead of throwing. E2B'scommands.run()throwsCommandExitErroron a non-zero exit, so a failing command rejected with no command or stderr instead of resolving with itsexitCode. The blueprint is bumped to version 2; runflue update sandbox e2bto update an existing integration (#805).
- New
- Fixed
Fix Claude models on the
cloudflare-ai-gatewayprovider failing with404 not_found_error: model: claude-sonnet-4.6 was not found. Since Pi 0.84.3 the gateway catalog lists Claude models with dotted versions (claude-sonnet-4.6), but the gateway's native Anthropic endpoint forwards the model ID to Anthropic, which only accepts dashed IDs. Flue now registers these models under Anthropic's IDs (cloudflare-ai-gateway/claude-sonnet-4-6); dotted specifiers still resolve to them. Thecloudflarebinding provider, which uses Cloudflare's dotted catalog names, is unchanged.
- Fixed
- Fixed
Fix model specifiers that stopped resolving after the Pi 0.87.1 upgrade. Pi renamed the
cloudflare-ai-gatewayClaude model IDs from dashed versions to dotted versions (claude-sonnet-4-6→claude-sonnet-4.6), so existing specifiers such ascloudflare-ai-gateway/claude-sonnet-4-6failed withUnknown model ID. When a model ID isn't in the catalog but matches exactly one catalog ID that differs only in-versus.between version digits, Flue now resolves to that model and logs a one-time warning suggesting the current ID.
- Fixed
- New
Add bounded conversation history reads so long-running chats no longer transfer the whole transcript on every page load or reconnect.
- New
Support document attachments (PDFs) on user messages.
DeliveredAttachmentis now a union of image and document attachments —{ type: 'document', data, mimeType: 'application/pdf', filename? }— accepted ondispatch(), theinit()handle, and direct HTTP prompts.session.prompt(),skill(), andtask()gain adocumentsoption alongsideimages, and@flue/react'ssendMessage()gains adocumentsoption. Documents are forwarded to the model as native document content: Anthropicdocumentblocks, OpenAI (and Azure OpenAI) Responsesinput_fileparts, and GoogleinlineData. On other model APIs the document is replaced in model context with a text placeholder saying it was omitted. Documents are stored like images — as canonical attachments projected asfileparts withmediaType: 'application/pdf'— so existing conversations and persistence adapters need no migration. The SDK also exportsDeliveredImageAttachmentandDeliveredDocumentAttachment. - New
Conversation messages now carry a server-authored
timestamp(ISO 8601 capture time of the underlying durable record) inhistory(),observe(), anduseFlueAgent. It works for every role and for existing conversations.metadataremains entirely agent-authored. Conversation settlements carry the sametimestampfor when the submission settled. - New
Adds support for GPT-6 Sol, GPT-6 Luna, and Claude Opus 5.5. Flue now uses Pi 0.87.1, so you can pass these models to
useModel()without extra configuration. The model list athttps://flueframework.com/models.jsonalso includes them now. There are no public API changes. - Fixed
Fix Cloudflare-routed Anthropic models (
cloudflare/anthropic/*) failing withbetas: Extra inputs are not permittedwhenever the agent had tools. Anthropic beta features are now sent in theanthropic-betarequest header instead of the request body.
- New
- New
Allow Flue packages and newly scaffolded applications to share a compatible Hono installation. Hono dependencies now use
^4.12.32, preventing fresh projects from installing a newer root Hono alongside the runtime's older exact version and failing typecheck. - Fixed
Clarify that a skill's
allowed-toolsfield is guidance. Flue accepts it for Agent Skills spec compatibility and does not enforce it; enforce authorization in your own tools and approval gates. - Fixed
Make trace content truncation scale linearly for long message arrays. Large traced conversations no longer repeatedly serialize the remaining history while fitting
gen_ai.input.messagesto the attribute budget, avoiding request latency and Cloudflare Durable Object CPU-limit resets. - Fixed
Record tool arguments and results of every payload shape on the standard
gen_ai.tool.call.arguments/gen_ai.tool.call.resultattributes, so OTel backends can display them. The shape-based diversion to the vendorflue.tool.call.*fallback keys is removed — those keys are no longer emitted, and their constants are retained for source compatibility. Payloads record exactly as before: objects/arrays as JSON strings, strings byte-for-byte within the content budget, other scalars as their JSON form.
- New
- New
Trace content budgets are now configurable.
createCloudflareTracing({ contentBudgetBytes })andcreateOpenTelemetryInstrumentation({ contentBudgetBytes })override the default 56 KiB per-span content pool — raise it (e.g.contentBudgetBytes: 200_000) to ship fuller prompts and tool results to an observability backend that isn't bound by workerd's 64 KiB span-attribute cap, or tighten it. The 128-byte sentinel floor and the shared-pool semantics are unchanged. - New
Tools returned by
createMcpConnection()now retain the metadata sent by the MCP server in itstools/listresponse. The metadata is available on each Flue tool astool.annotations: - New
Tools can now declare a
timeoutMsexecution bound: on expiry the harness aborts the tool'scontext.signaland settles the call with aToolTimeoutError(the model seesTool "<name>" timed out after <ms>msand the conversation continues) instead of letting one hung call consume the submission's durability budget. - Fixed
The OpenTelemetry ecosystem page and Cloudflare target guide now document
contentBudgetBytesoncreateOpenTelemetryInstrumentation()/createCloudflareTracing(): an override for the default 56 KiB per-span content pool. Raising it ships fuller content only on backends not bound by workerd's span cap (the OpenTelemetry adapter); on the Cloudflare target it is a tightening control only, since workerd's 64 KiB span-attribute cap is a platform limit no setting raises.
- New
- Fixed
Interrupted submissions no longer retry forever when the recovery render throws: once the submission's durability deadline passes, a submission whose classification render keeps failing is settled as timed out instead of being re-rendered on every supervisor wake without ever consuming an attempt.
- Fixed
The Cloudflare binding's Anthropic gateway path now maps the agent's
thinkingLevelto an adaptive-thinking effort (output_config.effort), souseModelthinking levels take effect on adaptive-thinking models instead of always running at Anthropic's default. - Fixed
The Cloudflare binding provider now accepts a
cacheRetentionoption ('short'or'long') to enable Anthropic prompt caching foranthropic/…gateway models — repeated prefixes are served from cache at the cached input rate instead of paying full input price every turn. Default'none'keeps the previous behavior. - Fixed
The Cloudflare Sandbox documentation page now presents
flue add sandbox cloudflareas a copyable prompt for your coding agent, with an explainer of what the blueprint-driven agent may do — installing@cloudflare/sandbox, wiring the Durable Object binding, migration, and containerDockerfile, and updating the agent to use the sandbox.
- Fixed
A whole project to check? Paste its package.json in the dependency check: every deprecated, outdated or vulnerable dependency, with what to upgrade to. Free, no account.