Bead Creator
Create a small, high-quality set of new Beads (“br” issues) for a repo based on quick repo understanding + a user-specified focus.
Quick start
Confirm repo target
- If the user gave a path, use that.
- Otherwise assume the current working directory is inside the repo.
- Determine the repo root by walking up until you find a
.git/ directory.
Confirm focus (unless the user already specified)
Ask a single question:
- “What should these beads optimize for: reliability, new features, performance, security, marketing, docs, refactor/tech-debt, or something else?”
Ensure Beads is initialized
- If
{repoRoot}/.beads/ does not exist: run br init from {repoRoot}.
Fast repo understanding
- Read (if present):
README.md, ARCHITECTURE.md.
- Build a lightweight mental model of the repo using quick searches (don’t deep-read everything).
- Do not create new documentation files or “repo map” artifacts; keep any notes in the chat response only.
Create 3–5 beads
- Keep each bead small and action-oriented.
- Use
br create ... --json and capture the IDs.
- Add labels (optional but recommended) to encode focus/area.
Report back
- Post a short list: bead id → title → why it matters → “next step” / verification hint.
Repo scan checklist (lightweight)
Run these from {repoRoot} (use --no-color when piping).
Core docs
README.md (how to run/test/deploy; stated goals)
ARCHITECTURE.md (components/boundaries; operational constraints)
Structure / hotspots
- Top-level layout:
ls -la
- Language/build clues:
package.json, pnpm-lock.yaml, yarn.lock, tsconfig.json
pyproject.toml, requirements.txt
Cargo.toml, go.mod
Dockerfile, docker-compose.yml
.github/workflows/*, Makefile, Justfile
Quick search targets
- Work debt markers:
rg -n "TODO|FIXME|HACK|XXX" .
- Stability markers:
rg -n "retry|backoff|timeout|circuit|rate limit" .
- Observability markers:
rg -n "log(ger)?\\.|metrics|tracing|sentry|otel|prometheus" .
- Tests:
find . -maxdepth 4 -type d \( -name test -o -name tests -o -name __tests__ \)
Bead writing rules
1) Prefer small, parallelizable work
Each bead should be roughly 30–120 minutes and have minimal shared surface area.
2) Include acceptance criteria
A bead without acceptance criteria becomes a conversation, not a task.
3) Avoid speculative refactors
Unless the focus explicitly asks for refactors, prefer beads that have a clear “why now”.
Bead template (use this in --description)
Use a compact structure:
- Goal:
- Context: (files/dirs; what you observed)
- Approach: (high level)
- Acceptance: (bullet list)
- Verify: (command(s) or observable behavior)
Example description:
Goal: Improve reliability of under transient failures.
Context: Found <file/path> with and no timeout/retry.
Approach: Add timeout + bounded retries with jitter; surface errors clearly.
Acceptance:
- Requests time out after s
- Retries capped at
- Error is logged with request id
Verify:
- Run
- Simulate failure by
Creating beads (commands)
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
Focus guidance
If the user specifies a focus, bias bead selection accordingly:
- Reliability: timeouts/retries, idempotency, backoff, failure modes, CI stability, error handling
- Performance: obvious hotspots, N+1 patterns, caching, build performance, large bundle size
- Security: secrets handling, authz checks, dependency hygiene, SSRF/path traversal hazards
- New features: “thin vertical slice” improvements that fit existing architecture
- Docs: missing runbooks, setup steps, troubleshooting, architecture diagrams, onboarding
- Marketing: landing page copy debt, SEO pages, product docs that convert (only if repo is marketing-related)
- Refactor/tech debt: simplify modules, remove duplication, improve boundaries, add types/tests
Guardrails
- Do not change repo code unless the user asked for implementation; this skill’s job is to create beads.
- Do not create new documentation files (no README/ARCHITECTURE updates, no new “repo-map” docs). Any repo understanding stays in the chat response.
- Only touch
.beads/ (via br).
- Do not run git commands unless the user explicitly asks.
- If repo context is insufficient, ask 1–2 targeted questions rather than guessing.
Converted and distributed by TomeVault — claim your Tome and manage your conversions.
1---2name: grp06-useful-codex-skills-bead-creator3description: Bead Creator4---56# Bead Creator78Create a small, high-quality set of new Beads (“br” issues) for a repo based on quick repo understanding + a user-specified focus.910## Quick start11121. **Confirm repo target**13 - If the user gave a path, use that.14 - Otherwise assume the current working directory is inside the repo.15 - Determine the repo root by walking up until you find a `.git/` directory.16172. **Confirm focus** (unless the user already specified)18 Ask a single question:19 - “What should these beads optimize for: reliability, new features, performance, security, marketing, docs, refactor/tech-debt, or something else?”20213. **Ensure Beads is initialized**22 - If `{repoRoot}/.beads/` does not exist: run `br init` from `{repoRoot}`.23244. **Fast repo understanding**25 - Read (if present): `README.md`, `ARCHITECTURE.md`.26 - Build a lightweight *mental model* of the repo using quick searches (don’t deep-read everything).27 - Do **not** create new documentation files or “repo map” artifacts; keep any notes in the chat response only.28295. **Create 3–5 beads**30 - Keep each bead small and action-oriented.31 - Use `br create ... --json` and capture the IDs.32 - Add labels (optional but recommended) to encode focus/area.33346. **Report back**35 - Post a short list: bead id → title → why it matters → “next step” / verification hint.3637## Repo scan checklist (lightweight)3839Run these from `{repoRoot}` (use `--no-color` when piping).4041### Core docs42- `README.md` (how to run/test/deploy; stated goals)43- `ARCHITECTURE.md` (components/boundaries; operational constraints)4445### Structure / hotspots46- Top-level layout: `ls -la`47- Language/build clues:48 - `package.json`, `pnpm-lock.yaml`, `yarn.lock`, `tsconfig.json`49 - `pyproject.toml`, `requirements.txt`50 - `Cargo.toml`, `go.mod`51 - `Dockerfile`, `docker-compose.yml`52 - `.github/workflows/*`, `Makefile`, `Justfile`5354### Quick search targets55- Work debt markers:56 - `rg -n "TODO|FIXME|HACK|XXX" .`57- Stability markers:58 - `rg -n "retry|backoff|timeout|circuit|rate limit" .`59- Observability markers:60 - `rg -n "log(ger)?\\.|metrics|tracing|sentry|otel|prometheus" .`61- Tests:62 - `find . -maxdepth 4 -type d \( -name test -o -name tests -o -name __tests__ \)`6364## Bead writing rules6566### 1) Prefer small, parallelizable work67Each bead should be roughly **30–120 minutes** and have minimal shared surface area.6869### 2) Include acceptance criteria70A bead without acceptance criteria becomes a conversation, not a task.7172### 3) Avoid speculative refactors73Unless the focus explicitly asks for refactors, prefer beads that have a clear “why now”.7475## Bead template (use this in `--description`)7677Use a compact structure:7879- **Goal:**80- **Context:** (files/dirs; what you observed)81- **Approach:** (high level)82- **Acceptance:** (bullet list)83- **Verify:** (command(s) or observable behavior)8485Example description:8687Goal: Improve reliability of <component> under transient failures.88Context: Found <file/path> with <behavior> and no timeout/retry.89Approach: Add timeout + bounded retries with jitter; surface errors clearly.90Acceptance:91- Requests time out after <N>s92- Retries capped at <M>93- Error is logged with request id94Verify:95- Run <command>96- Simulate failure by <method>9798## Creating beads (commands)99100Create 3–5 beads. Prefer `task` unless it’s clearly a `bug` or `feature`.101102```bash103br create "<title>" --type task --priority 2 --description "<template>" --json104```105106Optional labels (use consistent tokens):107```bash108br label add <id> focus:reliability area:observability109```110111## Focus guidance112113If the user specifies a focus, bias bead selection accordingly:114115- **Reliability:** timeouts/retries, idempotency, backoff, failure modes, CI stability, error handling116- **Performance:** obvious hotspots, N+1 patterns, caching, build performance, large bundle size117- **Security:** secrets handling, authz checks, dependency hygiene, SSRF/path traversal hazards118- **New features:** “thin vertical slice” improvements that fit existing architecture119- **Docs:** missing runbooks, setup steps, troubleshooting, architecture diagrams, onboarding120- **Marketing:** landing page copy debt, SEO pages, product docs that convert (only if repo is marketing-related)121- **Refactor/tech debt:** simplify modules, remove duplication, improve boundaries, add types/tests122123## Guardrails124125- **Do not change repo code** unless the user asked for implementation; this skill’s job is to create beads.126- **Do not create new documentation files** (no README/ARCHITECTURE updates, no new “repo-map” docs). Any repo understanding stays in the chat response.127- **Only touch `.beads/`** (via `br`).128- **Do not run git commands** unless the user explicitly asks.129- If repo context is insufficient, ask 1–2 targeted questions rather than guessing.130131---132> Converted and distributed by [TomeVault](https://tomevault.io/claim/grp06) — claim your Tome and manage your conversions.133<!-- tomevault:4.0:skill_md:2026-04-11 -->