You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The user asks you to implement a new feature. You respond in multiple steps:
29
29
30
-
1. Spawn two different file-picker-max's with different prompts to find relevant files; spawn a code-searcher and glob-matcher to find more relevant files and answer questions about the codebase; spawn 1 docs researcher to find relevant docs.
31
-
1a. Read all the relevant files using the read_files tool.
32
-
2. Spawn one more file-picker-max and one more code-searcher with different prompts to find relevant files.
33
-
2a. Read all the relevant files using the read_files tool.
34
-
3. Use the str_replace or write_file tool to make the changes.
35
-
4. Test your changes by running appropriate validation commands for the project (e.g. typechecks, tests, lints, etc.). You may have to explore the project to find the appropriate commands.
36
-
5. End your turn.`,
30
+
- Gather context on the user's request
31
+
- Use the write_todos tool to write out your step-by-step implementation plan.
32
+
- Use the str_replace or write_file tool to make the changes.
33
+
- Test your changes by running appropriate validation commands for the project (e.g. typechecks, tests, lints, etc.). You may have to explore the project to find the appropriate commands.
@@ -82,8 +75,6 @@ Continue to spawn layers of agents until have completed the user's request or re
82
75
- **Sequence agents properly:** Keep in mind dependencies when spawning different agents. Don't spawn agents in parallel that depend on each other. Be conservative sequencing agents so they can build on each other's insights:
83
76
- Spawn file pickers, code-searcher, directory-lister, glob-matcher, commanders, and researchers before making edits.
84
77
${buildArray(
85
-
hasDecomposingThinker&&
86
-
'- Spawn a decomposing-thinker agent after you have gathered all the context to ask key questions and plan your response. Make sure to include all the relevant file paths in the params.',
87
78
isMax&&
88
79
'- Spawn a base2-gpt-5-worker agent inline after you have gathered all the context you need (and not before!).',
89
80
).join('\n ')}
@@ -152,13 +143,7 @@ ${buildArray(
152
143
'- You must spawn a file-researcher to find relevant files; consider also spawning a web and/or docs researcher to find relevant information online.',
153
144
'- Read **ALL** the files that the file-researcher found using the read_files tool. It is important that you read every single file that the file-researcher found. This is the only time you should use read_files on a long list of files -- it is expensive to do this more than once!',
154
145
`- Consider spawning other agents or reading more files as needed to gather comprehensive context to answer the user's request.`,
155
-
hasDecomposingThinker&&
156
-
`- Spawn a decomposing-thinker agent to ask key questions and plan your response. Make sure to include all the relevant file paths in the params.`,
157
-
usesTodos&&
158
-
`- Use the write_todos tool to write out your step-by-step implementation plan.${hasNoValidation ? '' : ' You should include at least one step to validate/test your changes.'}`,
159
-
isFast&&
160
-
!usesTodos&&
161
-
'- Write out your implementation plan as a bullet point list.',
146
+
isFast&&`- Use the write_todos tool to write out your step-by-step implementation plan.${hasNoValidation ? '' : ' You should include at least one step to validate/test your changes.'}`,
162
147
isFast&&
163
148
`- Use the str_replace or write_file tool to make the changes. (Pause after making all the changes to see the tool results of your edits and double check they went through correctly.)`,
0 commit comments