🤖 fix: simplify plan mode instructions#2840
Conversation
|
@codex review |
|
Codex Review: Didn't find any major issues. Can't wait for the next one! ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
|
@codex review |
ammar-agent
force-pushed
the
fix/plan-mode-instructions
branch
from
8159b13 to
3f713b4
Compare
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3f713b43be
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Move the detailed planning guidance into the built-in plan agent so custom plan agents can override it, while keeping runtime injection focused on workspace-specific plan-file constraints. Also simplify the plan instructions and tighten tests around behavior instead of exact prompt prose. --- _Generated with `mux` • Model: `openai:gpt-5.4` • Thinking: `xhigh` • Cost: `$44.79`_ <!-- mux-attribution: model=openai:gpt-5.4 thinking=xhigh costs=44.79 -->
ammar-agent
force-pushed
the
fix/plan-mode-instructions
branch
from
3f713b4 to
dee3ebb
Compare
|
@codex review Please take another look. |
|
Codex Review: Didn't find any major issues. Already looking forward to the next diff. ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
2 participants
Summary
Move the opinionated plan-mode guidance into the built-in
planagent body so custom plan agents can override it, while leaving runtime injection focused on workspace-specific plan-file constraints.Background
Plan guidance was split between the built-in plan spec and runtime-injected
modeUtilstext. That made the behavior harder to override and left the instructions more repetitive than necessary.Implementation
getPlanModeInstruction()down to runtime-only facts: exact plan path, plan-file lifecycle guidance, edit restrictions, and read-only bash limitsplanagent instructions while keeping the key Explore /ask_user_question/propose_planworkflow expectationsValidation
make static-checkbun test src/common/utils/ui/modeUtils.test.ts src/node/services/streamContextBuilder.test.ts src/node/services/agentDefinitions/agentDefinitionsService.test.tsRisks
Low. This changes prompt composition for plan-like agents plus docs/tests. The main regression risk is accidentally dropping plan-mode guardrails or prompt ordering, which is covered by the updated prompt assembly and inheritance tests.
Generated with
mux• Model:openai:gpt-5.4• Thinking:xhigh• Cost:$44.79