bead-creator/SKILL.md
Analyze a code repository (README.md, ARCHITECTURE.md, and lightweight codebase search) and create 3–5 new Beads issues via br (beads_rust). Use when the user asks to create/generate beads, seed a backlog, open tasks/issues, or wants 3–5 task ideas focused on reliability, new features, performance, security, refactors, docs, or marketing for a specific repo.
npx skillsauth add grp06/useful-codex-skills bead-creatorInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Create a small, high-quality set of new Beads (“br” issues) for a repo based on quick repo understanding + a user-specified focus.
Confirm repo target
.git/ directory.Confirm focus (unless the user already specified) Ask a single question:
Ensure Beads is initialized
{repoRoot}/.beads/ does not exist: run br init from {repoRoot}.Fast repo understanding
README.md, ARCHITECTURE.md.Create 3–5 beads
br create ... --json and capture the IDs.Report back
Run these from {repoRoot} (use --no-color when piping).
README.md (how to run/test/deploy; stated goals)ARCHITECTURE.md (components/boundaries; operational constraints)ls -lapackage.json, pnpm-lock.yaml, yarn.lock, tsconfig.jsonpyproject.toml, requirements.txtCargo.toml, go.modDockerfile, docker-compose.yml.github/workflows/*, Makefile, Justfilerg -n "TODO|FIXME|HACK|XXX" .rg -n "retry|backoff|timeout|circuit|rate limit" .rg -n "log(ger)?\\.|metrics|tracing|sentry|otel|prometheus" .find . -maxdepth 4 -type d \( -name test -o -name tests -o -name __tests__ \)Each bead should be roughly 30–120 minutes and have minimal shared surface area.
A bead without acceptance criteria becomes a conversation, not a task.
Unless the focus explicitly asks for refactors, prefer beads that have a clear “why now”.
--description)Use a compact structure:
Example description:
Goal: Improve reliability of <component> under transient failures. Context: Found <file/path> with <behavior> and no timeout/retry. Approach: Add timeout + bounded retries with jitter; surface errors clearly. Acceptance:
Create 3–5 beads. Prefer task unless it’s clearly a bug or feature.
br create "<title>" --type task --priority 2 --description "<template>" --json
Optional labels (use consistent tokens):
br label add <id> focus:reliability area:observability
If the user specifies a focus, bias bead selection accordingly:
.beads/ (via br).development
Create an ExecPlan from a locked refactor decision, PRD, RFC, or detailed problem statement, following the repo's PLANS.md. Use when the user asks for an exec plan, execution plan, or ExecPlan, or wants a decided refactor turned into a step-by-step plan.
documentation
Update ARCHITECTURE.md after implementing an execution plan. Use when user asks to update architecture docs, sync architecture, refresh architecture after implementation, or mentions updating docs post-execplan.
tools
Analyze Codex rollout transcripts for one repository and time window, extract repo-scoped tool calls that failed, cluster repeated error patterns, and recommend the next agent/tooling improvements. Use when the user wants to inspect tool-call failures, operational friction, repeated command mistakes, patch failures, or prompt/tooling improvements based on recent Codex runs.
development
Analyze Codex session logs for one repository, extract the repo-specific top-level conversations for a target day, summarize every user message in order, and surface the next best action. Use when the user asks what happened yesterday in a repo, which conversations touched a repo, what to work on next based on recent Codex work, or to reconstruct repo workstreams from session history.