fix: use instructions for system prompt and prompt for user message in subagents#6121
Merged
tlongwell-block merged 2 commits intomainfrom Dec 16, 2025
Merged
fix: use instructions for system prompt and prompt for user message in subagents#6121tlongwell-block merged 2 commits intomainfrom
tlongwell-block merged 2 commits intomainfrom
Conversation
Collaborator
Author
|
/goose give me a minimal review on this change |
Contributor
SummaryThis PR correctly separates 🔴 Blocking IssuesNone found. 🟡 WarningsNone found. 🟢 SuggestionsNone. ✅ Highlights
Review generated by goose |
DOsinga
approved these changes
Dec 16, 2025
Collaborator
DOsinga
left a comment
There was a problem hiding this comment.
I think goose missed this?
katzdave
approved these changes
Dec 16, 2025
zanesq
added a commit
that referenced
this pull request
Dec 16, 2025
…s-predefined-models * 'main' of github.com:block/goose: (81 commits) fix: display shell output as static text instead of spinner (#6041) fix : Custom providers with empty API keys show as configured in desktop (#6105) Add .agents/skills and ~/.config/agent/skills to skills discovery paths (#6139) fix: use instructions for system prompt and prompt for user message in subagents (#6121) Fix compaction loop for small models or large input (#5803) feat: Centralize theme management with ThemeContext (#6137) OpenRouter & Xai streaming (#5873) fix: resolve mcp-hermit cleanup path expansion issue (#5953) feat: add goose PR reviewer workflow (#6124) perf: Avoid repeated MCP queries during streaming responses (#6138) Fix YAML serialization for recipes with special characters (#5796) Add more posthog analytics (privacy aware) (#6122) docs: add Sugar MCP server to extensions registry (#6077) Fix tokenState loading on new sessions (#6129) bump bedrock dep versions (#6090) Don't persist ephemeral extensions when resuming sessions (#5974) chore(deps): bump mdast-util-to-hast from 13.2.0 to 13.2.1 in /ui/desktop (#5939) chore(deps): bump node-forge from 1.3.1 to 1.3.2 in /documentation (#5898) Add Scorecard supply-chain security workflow (#5810) Don't show subagent tool when we're a subagent (#6125) ... # Conflicts: # crates/goose/src/providers/formats/databricks.rs
aharvard
added a commit
that referenced
this pull request
Dec 17, 2025
* main: fix: we don't need to warn about tool count when in code mode (#6149) deps: upgrade agent-client-protocol to 0.9.0 (#6109) fix(providers): fix for gemini-cli on windows to work around cmd's multiline prompt limitations #5911 (#5966) More slash commands (#5858) fix: MCP UI not rendering due to CallToolResult structure change (#6143) fix: display shell output as static text instead of spinner (#6041) fix : Custom providers with empty API keys show as configured in desktop (#6105) Add .agents/skills and ~/.config/agent/skills to skills discovery paths (#6139) fix: use instructions for system prompt and prompt for user message in subagents (#6121) Fix compaction loop for small models or large input (#5803) feat: Centralize theme management with ThemeContext (#6137) OpenRouter & Xai streaming (#5873) fix: resolve mcp-hermit cleanup path expansion issue (#5953) feat: add goose PR reviewer workflow (#6124) perf: Avoid repeated MCP queries during streaming responses (#6138) Fix YAML serialization for recipes with special characters (#5796) Add more posthog analytics (privacy aware) (#6122) docs: add Sugar MCP server to extensions registry (#6077)
dianed-square
added a commit
that referenced
this pull request
Dec 17, 2025
* origin/main: (57 commits) docs: create/edit recipe button (#6145) fix(google): Fix 400 Bad Request error with Gemini 3 thought signatures (#6035) fix: we don't need to warn about tool count when in code mode (#6149) deps: upgrade agent-client-protocol to 0.9.0 (#6109) fix(providers): fix for gemini-cli on windows to work around cmd's multiline prompt limitations #5911 (#5966) More slash commands (#5858) fix: MCP UI not rendering due to CallToolResult structure change (#6143) fix: display shell output as static text instead of spinner (#6041) fix : Custom providers with empty API keys show as configured in desktop (#6105) Add .agents/skills and ~/.config/agent/skills to skills discovery paths (#6139) fix: use instructions for system prompt and prompt for user message in subagents (#6121) Fix compaction loop for small models or large input (#5803) feat: Centralize theme management with ThemeContext (#6137) OpenRouter & Xai streaming (#5873) fix: resolve mcp-hermit cleanup path expansion issue (#5953) feat: add goose PR reviewer workflow (#6124) perf: Avoid repeated MCP queries during streaming responses (#6138) Fix YAML serialization for recipes with special characters (#5796) Add more posthog analytics (privacy aware) (#6122) docs: add Sugar MCP server to extensions registry (#6077) ...
katzdave
added a commit
that referenced
this pull request
Dec 17, 2025
…icing * 'main' of github.com:block/goose: (35 commits) docs: skills (#6062) fix: add conditional configuration for GOOSE_BIN_DIR in PATH (#5940) Update dependencies to help in Fedora packaging (#5835) fix: make goose reviewer less bad (#6154) docs: create/edit recipe button (#6145) fix(google): Fix 400 Bad Request error with Gemini 3 thought signatures (#6035) fix: we don't need to warn about tool count when in code mode (#6149) deps: upgrade agent-client-protocol to 0.9.0 (#6109) fix(providers): fix for gemini-cli on windows to work around cmd's multiline prompt limitations #5911 (#5966) More slash commands (#5858) fix: MCP UI not rendering due to CallToolResult structure change (#6143) fix: display shell output as static text instead of spinner (#6041) fix : Custom providers with empty API keys show as configured in desktop (#6105) Add .agents/skills and ~/.config/agent/skills to skills discovery paths (#6139) fix: use instructions for system prompt and prompt for user message in subagents (#6121) Fix compaction loop for small models or large input (#5803) feat: Centralize theme management with ThemeContext (#6137) OpenRouter & Xai streaming (#5873) fix: resolve mcp-hermit cleanup path expansion issue (#5953) feat: add goose PR reviewer workflow (#6124) ...
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes incorrect handling of
recipe.instructionsandrecipe.promptin the subagent system. Previously, both fields were merged together and used for BOTH the system prompt AND the first user message, which was redundant and semantically incorrect.The Bug
In
subagent_tool.rs, thebuild_subrecipe()function mergedpromptintoinstructions:Then in
subagent_handler.rs, this merged blob was used for both:task_instructions)Message::user().with_text(...))The Fix
instructions→ system prompt (agent personality/role/guidelines)prompt→ user message (the actual task to execute)This matches how the CLI handles recipes in
extract_from_cli.rs.Changes
subagent_tool.rssubagent_handler.rsinstructionsfor system prompt,promptfor user messagesubagent_system.mdBackward Compatibility
instructionsstill work (used for both, as before)promptstill work