v1.1.0
π DAIV v1.1.0 β Skills, Safer Prompts & Smarter Planning
DAIV v1.1.0 introduces an Agent Skills system, safer prompt templates, and better support for large repos and evaluations.
π Highlights
-
π§ Agent Skills system using Anthropic-style progressive disclosure.
-
π New skills:
creating-agents-md-fileβ generatesAGENTS.mdfollowing the AGENTS.md spec.maintaining-changelogβ maintains or creates changelogs (Keep a Changelog style).creating-daiv-yml-configβ helps author.daiv.ymlwith sandbox / formatting settings.
-
π Chunked file reading in
read_tool(start_line,max_lines, default 2000 lines) with clear truncation messages. -
π§ͺ SWERepoClient for SWE-benchβstyle runs on public OSS repos (no credentials, temp clones).
-
π€ Added model support:
gpt-5.1,gpt-5.1-codex,gpt-5.1-codex-mini. -
πΈ Anthropic caching middleware now supports OpenRouter.
π§ Agent & Middleware Changes
-
PlanAndExecuteAgent:- Default fallback planning model:
gpt-5.1. - Code review model:
gpt-5.1-codex-mini. - More structured, human-friendly plans with code minimalism guidelines.
- New
EXECUTION_THINKING_LEVELto enable thinking during execution. - Uses
ToolStrategyfor execution to better handle union types.
- Default fallback planning model:
-
Switched from toolkits to LangChain v1 middlewares:
FileNavigationMiddlewareFileEditingMiddlewareMergeRequestMiddlewareWebSearchMiddleware
-
Replaced
plan_think_toolwithTodoListMiddlewarefor explicit planning todos. -
All prompt templates migrated from Jinja2 β Mustache to mitigate code injection vectors.
-
New quick action:
/clone-to-topicto clone issues into all repos matching specified topics.
π Fixes
format_code_toolnow applies patches even when the format command fails..gitis no longer included in sandbox archives, reducing size and repo exposure.- Fixed
InvalidNamespaceErrorfor branch names with periods (e.g.fix/python-version-3.11) via sanitized LangGraph namespaces.