# Prd To Issues > Read an existing PRD (typically issues/prd.md, written by write-a-prd) and break it into vertical-slice tracer-bullet issues at issues/NNN-short-title.md, tagged HITL (needs human) or AFK (autonomous). Step 2 of the build pipeline (write-a-prd → prd-to-issues → work-issues). Use when a PRD exists and needs to become work tickets. NOT for writing the PRD itself (use write-a-prd). Triggers: /prd-to… - **Type:** Skill - **Install:** `agentstack add skill-belchman-claude-skills-prd-to-issues` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [belchman](https://agentstack.voostack.com/s/belchman) - **Installs:** 0 - **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [belchman](https://github.com/belchman) - **Source:** https://github.com/belchman/claude-skills/tree/master/plugins/agentic-engineering/skills/prd-to-issues ## Install ```sh agentstack add skill-belchman-claude-skills-prd-to-issues ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # PRD to Issues Break a PRD into independently-grabbable issues using vertical slices (tracer bullets), written as local markdown files. ## Process ### 1. Locate the PRD Ask for the PRD file path (e.g. `issues/prd.md`). If not in context, read it. ### 2. Explore the codebase (optional) If you haven't already, explore to understand current state. ### 3. Draft vertical slices Each issue is a **tracer bullet** — a thin vertical slice cutting through ALL layers end-to-end, NOT a horizontal slice of one layer. Slices are **HITL** (need human interaction — design review, architectural decision) or **AFK** (implementable and mergeable without it). Prefer AFK. - Each slice delivers a narrow but COMPLETE path through every layer (schema, API, UI, tests) - A completed slice is demoable or verifiable on its own - Prefer many thin slices over few thick ones ### 4. Quiz the user (or skip if AFK) Present a numbered list. For each slice: - **Title**: short descriptive name - **Type**: HITL / AFK - **Blocked by**: which slices (if any) must complete first - **User stories covered**: which from the PRD Ask: - Does the granularity feel right? (too coarse / too fine) - Are dependencies correct? - Should any slices be merged or split? - Are HITL/AFK labels right? Iterate until approved. **AFK fallback** (no user available — `/feature` running headlessly, `claude -p` invocation, etc.): skip the quiz. Make your best granularity call from the PRD alone, then proceed straight to step 5. In each issue file's `## What to build` section, add a `### Slice rationale` sub-section noting why this slice exists as its own ticket (vs. being folded into a sibling) — a future reviewer or downstream agent reads it to challenge the slicing at the next checkpoint. If you're uncertain between two slicings, write a sibling `QUESTIONS.md` listing the choice you made and the alternative. **Narrow-feature note:** The "vertical slice through all layers (schema, API, UI, tests)" framing assumes a multi-layer feature. For a single-file CLI flag, a one-script change, or a docs-only update, the layers may collapse to one. Don't force-split a one-issue feature into three issues just to satisfy "many thin slices"; one issue with clear acceptance criteria is the right answer when there's only one layer to cut through. The most common natural 2-issue shapes for narrow features: - **Deep module + wiring** — one issue for the testable helper (no CLI noise), one for the CLI-level integration. Pattern: implementing `archive_cache(src, dest) -> sha_hex` separately from `--backup-cache ` flag wiring. - **Code + discovery doc** — for plugin/skill/library features where a SKILL.md or README discovery surface needs to advertise the new behavior. The doc lives in a different file (often a different lane — `### frontend` in this repo's split), and reviewers care about different things (correctness vs. discoverability/example accuracy). Pattern: `--json` flag in `crap.py` (issue 001) + `## Invocation` update in `crap/SKILL.md` (issue 002). **Blocking direction:** when one issue depends on another, prefer **doc-blocked-on-code** (not the reverse). The doc should cite the as-shipped flag spelling, exit code, output format, etc. — landing the doc first means it documents a spec, not behavior, and tends to drift. Exception: when the doc IS the spec contract (e.g. an OpenAPI schema that drives codegen), invert. ### 5. Create the issue files For each approved slice, write `issues/NNN-short-title.md` (e.g. `issues/001-add-user-auth.md`). Number from the next available slot (check existing files in `issues/`). Create files in dependency order (blockers first) so "Blocked by" can reference real filenames. Do NOT use `gh issue create` or any GitHub CLI. Do NOT reference GitHub issue numbers. Local filenames only for cross-references. ## Parent PRD `issues/prd.md` (or whichever PRD was used) ## Type **HITL** or **AFK** — pick one. HITL means a human needs to be in the loop (design review, credentials, architectural call). AFK means it can be picked up and merged autonomously. ## What to build Concise description of this vertical slice. End-to-end behavior, not layer-by-layer implementation. Reference sections of the parent PRD rather than duplicating. ## Acceptance criteria - [ ] Criterion 1 - [ ] Criterion 2 - [ ] Criterion 3 ## Blocked by - Blocked by `issues/NNN-title.md` (if any) Or "None - can start immediately". ## User stories addressed Reference by number from the parent PRD: - User story 3 - User story 7 Do NOT close or modify the parent PRD file. ## Next After issues are written, optionally run `write-a-rubric` per AFK issue to define grader-checkable success criteria as a sibling `issues/NNN-*.rubric.md`. Then run `work-issues` to start working the queue. ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [belchman](https://github.com/belchman) - **Source:** [belchman/claude-skills](https://github.com/belchman/claude-skills) - **License:** MIT Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-belchman-claude-skills-prd-to-issues - Seller: https://agentstack.voostack.com/s/belchman - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.