With compaction.inContext enabled, compaction can generate the summary inside the cached conversation: it repeats the request the next turn would send, appends the instructions as a system message plus a user message, and sets toolChoice none. The recent tail is kept as before. A gate requires a verified model, a warm cache from the same model and thinking level, room in the context window, and a lower estimated cost than the standalone summary. Failed or rejected attempts fall back to the standalone summary, never execute tools, and are recorded in details.inContext with their usage.
pi-ai adds the supportsInContextCompaction compat setting and test/in-context-compaction-probe.ts, which measures whether a model keeps its prompt cache for this request shape. The generated catalog flags the models that passed.
Increase estimated input size to leave more room when calculating
output limits. Verified against OpenRouter DeepSeek V4 Flash: the
same large input now succeeds with an explicit output limit.
Fixes#10497
runtime.context() keeps the range its last read scanned for the task
invocation. A later read with the same conversation and head marker scans
only entries committed after that range's tail and re-derives the view in
memory. A generation's request no longer rereads the whole transcript after
its prepare. The range is dropped when the invocation ends or sleeps.
Selection coordinates survived session switches and other transcript rebuilds, highlighting unrelated text in the new transcript. Reset selection and multi-click state before re-rendering session entries.
Based on #9310.
fixes#9311
Co-authored-by: xdagiz <xdagiz@protonmail.com>
A --tools list of only +name/-name entries changes the default tool
selection like the defaultTools setting, e.g. pi -t +codemode. Mixed
lists and patterns in modifiers are rejected, and /reload keeps tools
removed with -name removed.
StreamOptions.headers documents that caller values override provider
defaults, and every other provider applies model/caller headers after its
User-Agent default. The Codex provider set originator and User-Agent after
merging caller headers, so they could not be overridden. Set the defaults
first; Authorization and chatgpt-account-id stay fixed.
Co-authored-by: Lucas Meijer <308974+lucasmeijer@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: David Brailovsky <davidbrai@gmail.com>
This adds llama.cpp support for classifiers. Classifier models (Julia-1, Laya, Kev, lev, OpenJev) are listed as typesafe-system-one classifiers instead of chat models; other models keep the llama-cpp-classify fallback. Only 501 and 404 mark a model as chat, and classifier models restored from the model store are checked again once loaded.
Rules and the tool list follow the declared tools (active minus
hiddenDeclarations). The skills hint names no tool when the reader is
hidden. Codemode shows each tool's prompt guidelines with its
declaration, via the new ToolLoadout.getPromptGuidelines().
closes#10343
close() now aborts a per-connection signal that closes the in-flight
client, cancels the retry wait, and waits for the connect attempt to
settle before returning.
closes#10249
Freeze built-ins and make built-in globals read-only before the script
runs, keeping instance overrides working via accessors. Validate all
worker payloads on the host and fail malformed ones as sandbox errors
instead of crashing the host process.
closes#10444
--tools no longer removes MCP tools unless an entry starts with mcp__, so
pi --tools codemode can reach MCP servers. Unnamed MCP tools are never
declared directly; only tool_search can load them. --no-tools still
removes everything.
--tools and --exclude-tools accept * patterns, for example
--tools read,codemode,'mcp__radius__*'. --no-mcp disables the built-in
MCP support for one run.
OpenID Connect authorization servers treat a registration without application_type as a web client and reject http loopback redirect URIs. Derive it from redirect_uris when unset (MCP SEP-837): native for loopback hosts and custom schemes, otherwise web.
closes#10493
Bedrock's HTTP/2 connection can stall until the SDK's 5-minute session
timeout, which fails the request with "The pending stream has been
canceled" (ERR_HTTP2_STREAM_CANCEL). The request was never sent, so it
is safe to retry, but the message was only retried when a DNS cause was
appended.
closes#10379
- connectSsh detects the remote system, deploys the packaged daemon
verified by SHA-256 before it runs, and connects over ssh with strict
host keys under a fixed alias; scanHostKey and acceptHostKey
- Windows detection and deployment through PowerShell
- release builds the daemon for every supported remote system and
ships them in @earendil-works/pi-env's bin/
- end-to-end tests against a disposable sshd
- RemoteExecutionEnv: a Durable ExecutionEnv on another machine whose
results match NodeExecutionEnv there; Connection to the pi-env daemon
- packages/env/daemon: the Rust daemon for Linux, macOS, Android and
Windows, following Node and libuv semantics
- env.yml runs Durable's env tests, the daemon's checks and pi-env's
conformance and differential tests on Linux, macOS and Windows
* fix(coding-agent): keep syntax colors on multiline tokens
- format each line of a highlighted span separately
- add regression test for python docstrings
- fixes#10143
* fix(coding-agent): use text color for string interpolations
- map highlight.js `subst` scope to text color
