Skip to content

Caudra 0.2 Preview 10

View Markdown~3,380 tokens

v0.2.0-preview.10 on GitHub

  • One code tool replaces mcp and python_execution, and can ask the decision engine to judge and rerank results.
  • Search past conversations, from the code tool and from the TUI with /search-history.
  • Two source code tools instead of five, renamed source_code_context and source_code_refs, that every model sees and that list call sites.
  • Decision models in /login and /model, Auto mode out of experimental, and workflow decide() in the OpenAI Decisions question shape.
  • Pictures in the transcript.
  • Notes on your answers to the agent's questions, and drag to copy in /btw and the question form.
  • Goals stop when they make no progress, wait for CI or a deploy, and tell automations when they are stuck.
  • Subscription image generation works again after GPT-5.5 left Codex.

Important

Close every other Caudra session before you start Preview 10. On its first start, Preview 10 upgrades the session database in your state directory and needs exclusive access, including from editor sessions over ACP and caudra storage commands. While any of them has the database open, Preview 10 lists them and does not start, and nothing is changed.

The upgrade keeps your sessions and saves the old database as caudra.db.v28.bak next to caudra.db. Preview 9 and earlier cannot open the upgraded database, and caudra rollback restores only the previous binary, not the database.

Warning

Before you start Preview 10, rename ui.tool_output_lines.python_execution to ui.tool_output_lines.code. A caudra.toml that still sets the old key fails to load, with an error that names the new one. Permission rules, agent.disabled_tools entries, and system prompt profiles that name the mcp or python_execution tool no longer match anything, so name code in them instead. See Code.

Warning

If you run a remote Workcell server, update it with Caudra and rename its python_execution tool group to code. --tool-group python_execution and python_execution in WORKCELL_MCP_TOOL_GROUPS fail at startup, and Caudra refuses a server that does not serve code. See Remote workspaces.

Warning

If your workflows call decide(), rewrite their questions when you update. Questions are now a list, and each names itself with name. noul questions become predicate questions, and criteria becomes choices for a choice and levels for a score. Answers come back as answers, a list in question order, and by_name, keyed by name, and a predicate's probability is in probability. A workflow that passes the old shape fails at its first decide() call with a catchable error, and so does a run started before the update when you resume it, even at a call it had already answered. See Typed decisions.

Warning

If your config, profiles, or remote Workcell use the code-graph tools, update them together with Caudra. code_context and code_refs are now source_code_context and source_code_refs. code_map, code_impact, and code_expand are gone: use source_code_context for code_map or code_expand, and source_code_refs with depth for code_impact. The profile tool group code_graph is now source_code, [decisions.features] code_context is now source_code_context, and an old name in a config or profile is rejected as unknown. A remote Workcell must run the same release as Caudra, with --tool-group source_code instead of --tool-group code_graph. The SDK's tool kinds are now SourceCodeContext and SourceCodeRefs. Your sessions, /tools usage, decision log, and saved permission rules move to the new names on first start. See Code Intelligence.

Warning

Before you start Preview 10, move your decision endpoint out of caudra.toml. The [decisions] table now holds only the settings every endpoint shares. base_url, protocol, model, api_key_env, allow_remote, and allow_http in it, or in caudra.setup({ decisions = ... }), stop Caudra with an error that shows the table to write instead. To migrate:

  1. Run caudra auth login --decision, or move those keys into a [decisions.<name>] table in the global providers.toml, such as [decisions.typesafe]. Then delete them from [decisions].
  2. Delete allow_remote. Configuring an endpoint is now the consent to send decision context to it. Keep allow_http = true only on an endpoint that uses plain HTTP to a host other than numeric loopback.
  3. Delete decision_engine from [experimental]. It is now an unknown key.
  4. If you set TYPESAFE_BASE_URL or OPENAI_BASE_URL to move the decision endpoint, write base_url = "{env:TYPESAFE_BASE_URL}" in its table instead. Neither variable changes a decision endpoint any more.
  5. Run caudra decisions use <name> for a table you wrote by hand. Login makes a new endpoint active, but a hand-written one stays inactive until you select it.

See Decision endpoints.

Important

Intel macOS users building with Nix must switch to the native build. Current nixpkgs no longer supports Intel macOS, so the flake no longer provides x86_64-darwin outputs. Native Intel macOS builds and release binaries remain supported.

  • One code tool replaces mcp and python_execution. The model runs all of its Python in code: computation, MCP server calls, history search, and the decision helpers. It is offered in every session, with or without MCP servers, and its definition never changes, so the prompt cache holds. Python always runs on your machine, also in remote Workcell sessions, with 30 seconds of computation per run where python_execution stopped at 5 seconds, and the clock stops while a snippet waits for a call. timeoutSec is gone. Variables persist between runs, and reset: true starts a run empty. /tools, caudra tools, and /mcp mark MCP tools via code, and a session with an mcp card from Preview 9 still opens. See Code and Where snippets run.
  • Search your past conversations. history.search() finds what you wrote, what the model answered and reasoned, and what its tools did, in this conversation, this project, another project, or every stored session, compacted turns and subagents included. history.read() opens a hit in full, and history.sessions() lists sessions. Searching this project never asks, and another project or all projects ask first. With [decisions.features] history_search = "enforce", the decision engine reranks the matches and returns the 10 most relevant. See Search past conversations and Ranking.
  • Decisions and reranking in snippets. With a decision endpoint active, a snippet can call decide() for typed answers about its data, and rerank() to keep only the candidates relevant to a query, for up to 256 candidates, so the rest never reach the model. Calls need no approval and work in plan mode. Each shows as a row on the code card with a bar per option and a warning when the winner is uncertain, and /decisions draws logged answers the same way. See Decisions and reranking.
  • Decision models in /login and /model. /login has a Decision models section with TypeSafe Jev, OpenAI Luna, and a custom endpoint, and a new endpoint becomes the active one. In /model, the Decisions row of the Jobs section switches endpoints, turns decisions off, or adds one, and /decision-model opens it directly. /decision-next or Ctrl+X d moves to the next endpoint. While an endpoint is active, the status bar shows [decisions: <name>]. See Decision endpoints.
  • Decision endpoints from the command line. caudra auth login --decision and caudra auth logout --decision <name> add and remove an endpoint and its key, and caudra decisions list, use <name|off>, and status show and switch them. Without a saved key, Caudra reads the variable that the endpoint's api_key_env names, such as TYPESAFE_API_KEY or OPENAI_API_KEY. See CLI.
  • Auto mode and the decision engine are no longer experimental. /auto, --auto, SDK --permission-mode auto, /decisions, and caudra decisions work without a switch, and always_auto = true and sessions saved in Auto start in Auto again. With no active endpoint, Auto runs the calls that deterministic checks clear and prompts for the rest. Every decision feature stays off until you turn it on. See Auto mode.
  • Workflow decide() takes OpenAI Decisions questions. Choice options and score levels can carry a description, and each answer lists the probability of every option or level.
  • caudra auth login keeps your comments in providers.toml. Login and logout replace only the table they change.
  • Pictures in the transcript. view_image and image_generate cards, Markdown images that point at local files, and prompts with attached images show the picture as a thumbnail, at full resolution in terminals with Kitty graphics and as coloured blocks elsewhere. Click a picture to open it at full size, then press o to open it in your system viewer, w to open it in the workbench, or y to copy its path. Markdown images are never fetched from the network. Set [ui] inline_images to "blocks", "sextants", "kitty", or "off" to choose. See Images.
  • Search past sessions. /search-history searches this project's saved sessions, and Ctrl+F still searches the current chat. Tab switches between This chat, Project, and All projects. Results show each match with its session and age and a preview, and words can carry kind:, tool:, since:, and until: filters. Enter opens the session and scrolls to the match. See Search past sessions.
  • Add a note to your answers. On a question the agent asks, F5 opens a note for anything the options leave out. It reaches the agent with your answers, or with the dismissal when you press Esc. The form's hints now start with Enter submit. See Keybindings.
  • Drag to copy in /btw and the question form. Dragging over a /btw thread copies the passage when you let go, the way the transcript copies. In the question form, a drag copies instead of picking the option it ends on, and Ctrl+C copies a selection before it stops the agent. See Commands.
  • A goal that stops getting anywhere stops its run. When three turns in a row make no progress, the run ends with the goal still active instead of spending continuations until the limit. Autopilot reviews the stall like a loop cap. Headless clients get a stalled goal event, and a goal_finished automation trigger that lists stalled in verdicts fires too. See Commands and Automations.
  • A goal waiting on something outside the session waits instead of stalling. When the goal waits on a change such as a CI run or a deploy, the run ends and the goal is checked again after a minute, then twice as long each time up to 30 minutes, for about two hours before it stalls. A message from you ends the wait, and a resumed session keeps it. Headless clients get a waiting goal event.
  • Two source code tools instead of five, under new names. source_code_context (formerly code_context) finds the code a task touches, and source_code_refs lists who calls a symbol and from which lines. With depth, source_code_refs shows what a change could reach and how many of those symbols are tests, which is what code_impact did. Both tools are always in the request for every model.
  • More precise source code results. Standard library calls such as .clone() or Vec::new() no longer link to a symbol of the same name in your tree, and source_code_context lifts only names written as code, so an ordinary word such as "renew" no longer pulls in every new.
  • /tools counts calls made inside batch. Each call a batch makes has its own row.
  • ChatGPT subscription image generation uses a supported model. Image requests now run through GPT-6.1 Sol instead of GPT-5.5, which OpenAI retired from Codex with ChatGPT sign-in on October 14, without changing your configured image model or switching to API billing.
  • Lua job cleanup. Releasing an abandoned jobwait no longer aborts the process.
  • Shell output filters. Bundled filters no longer match unrelated subcommands that start with a filtered name.
  • File indexes and code graphs. PHP clone arguments and partial application no longer report parse errors, and Go type aliases are included.
  • SVG source icons. An icon over the pixel budget is skipped, including through nested masks, clips, and filters.
  • HTML conversion. Code blocks keep blank lines and lazy-loaded images resolve.
  • Nix builds. Evaluation works without building the worker source first.

Website privacy and analytics