Add loop.sh and PROMPT files for beads-loop workflow#3
Conversation
Add beads-loop runner script and prompt templates (build, plan, review) from the prompts repository for standardized AI-assisted development workflows.
![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: eaec9fdcc7
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".
| 0c. For reference, each plugin lives in its own top-level directory (e.g. `productivity/`, `sales/`, `customer-support/`). See CLAUDE.md for the full repo structure. | ||
| 0a. Run `br ready --json --limit 1` to find the highest priority unblocked task. | ||
| 0b. Run `br show <task-id>` to read the full specification. | ||
| 0c. For reference, the application source code is in `src/*`. |
There was a problem hiding this comment.
This prompt now tells agents to look in src/*/src/lib/*, but this repository is organized as top-level plugin directories (see AGENTS.md repository structure section) and has no src tree. In build/plan mode, this misroutes code search and edits to non-existent paths, so tasks can be marked incomplete or implemented in the wrong place even when the required plugin files already exist.
Useful? React with 👍 / 👎.
| ULTIMATE GOAL: We want to achieve [project-specific goal]. Ensure all necessary tasks | ||
| exist as beads with proper dependencies so `br ready` always shows the right next work. |
There was a problem hiding this comment.
[project-specific goal] is an unresolved template token, so the planning prompt no longer communicates a concrete target for this repo. That removes the criterion planners use to decide whether epics/tasks are sufficient, leading to inconsistent or incomplete bead plans across runs.
Useful? React with 👍 / 👎.
|
@claude fix |
1 participant
Summary
loop.sh— beads-loop v3 production-grade loop runner for Claude CodePROMPT_build.md— build mode prompt for implementing ready beadsPROMPT_plan.md— plan mode prompt for analyzing and organizing beadsPROMPT_review.md— review mode prompt for code review against acceptance criteriaThese files provide standardized AI-assisted development workflows across all FashionUnited repositories.
🤖 Generated with Claude Code