to-tickets

Files (2)

• SKILL.md
• openai.yaml · agents/

SKILL.md (first lines)

# To Tickets

Break a plan, spec, or conversation into a set of **tickets** — tracer-bullet vertical slices, each declaring the tickets that **block** it.

The issue tracker and triage label vocabulary should have been provided to you — run `/setup-matt-pocock-skills` if not.

## Process

### 1. Gather context

Work from whatever is already in the conversation context. If the user passes a reference (a spec path, an issue number or URL) as an argument, fetch it and read its full body and comments.

### 2. Explore the codebase (optional)

If you have not already explored the codebase, do so to understand the current state of the code. Ticket titles and descriptions should use the project's domain glossary vocabulary, and respect ADRs in the area you're touching.

Look for opportunities to prefactor the code to make the implementation easier. "Make the change easy, then make the easy change."

### 3. Draft vertical slices

Break the work into **tracer bullet** tickets.

<vertical-slice-rules>

- Each slice cuts a narrow but COMPLETE path through every layer (schema, API, UI, tests) — vertical, NOT a horizontal slice of one layer
- A completed slice is demoable or verifiable on its own
- Each slice is sized to fit in a single fresh context window
- Any prefactoring should be done first

</vertical-slice-rules>

Give each ticket its **blocking edges** — the other tickets that must complete before it can start. A ticket with no blockers can start immediately.

Every ticket also declares its **out of scope** — what it must NOT do. Name each sibling ticket whose work borders this one (a feature another ticket owns that looks adjacent enough to be pulled in, a contract another ticket will finalize, a refactor another ticket will run). The agent executing this ticket must not reach for them. "Blocked by" says what gates this ticket; "out of scope" says what this ticket must not implement even if it looks easy. Both are required — the missing one is how a ticket silently absorbs its neighbours' work.

Draft each ticket's acceptance criteria as **executable assertions**, not wishes:

- Each positive criterion states the behaviour that must work and **how it is verified** — the seam it is tested at (from `to-spec`'s Testing Decisions) or the command that checks it. A ticket is done only when every criterion's verification passes.
- Every "must not" / "must never" in the spec gets a **negative criterion** with its own test: the assertion is the absence (the error type is X not Y, no sensitive value in any output, no undeclared fields or declarations). A spec sentence is weak; a test is strong — the agent works to make tests pass.
- Every criterion must be checkable by a test, a command, or an inspection script. If you cannot name the check, the criterion is a wish — rework it.
- Verification commands must run from the repository root using tooling the repository declares (package-manager scripts, repo config). Prefer repo-declared tooling over globally installed executables; when a global tool is unavoidable, the ticket must record its name, version, and installation/invocation method so another machine can reproduce it — a verification that only passes in your environment is not verification.

Copy every **global invariant** from the spec into every ticket whose code path touches it — secrets handling (no sensitive value in any output or log), path containment, stdout purity, error semantics, anything the spec marks "must never". An invariant is not owned by the ticket that finalizes its contract: it applies everywhere, and no ticket may temporarily violate it because another ticket will fix it later.

Each ticket that touches dependencies or tooling carries a **dependency/tooling guard**: new direct dependencies, checkers, configs, or repo-level workflows require explicit authorization in the ticket — without it they are out of scope; and any dependency change must own syncing the repo's context/instruction file (AGENTS.md, CLAUDE.md, …), the lockfile, and relevant ADRs.

**Wide refactors are the exception to vertical slicing.** A **wide refactor** is one mechanical change — rename a column, retype a shared symbol — whose **blast radius** fans across the whole codebase, so a single edit breaks thousands of call sites at once and no vertical slice can land green. Don't force it into a tracer bullet; sequence it as **expand–contract**. First expand: add the new form beside the old so nothing breaks. Then migrate the call sites over in batches sized by blast radius (per package, per directory), each batch its own ticket blocked by the expand, keeping CI green batch to batch because the old form still exists. Finally contract: delete the old form once no caller remains, in a ticket blocked by every migrate batch. When even the batches can't stay green alone, keep the sequence but let them share an integration branch that all block a final integrate-and-verify ticket — green is promised only there.

### 4. Quiz the user

Present the proposed breakdown as a numbered list. For each ticket, show:

- **Title**: short descriptive name
- **Blocked by**: which other tickets (if any) must complete first
- **What it delivers**: the end-to-end behaviour this ticket makes work

Ask the user:

- Does the granularity feel right? (too coarse / too fine)
- Are the blocking edges correct — does each ticket only depend on tickets that genuinely gate it?
- Are the out-of-scope boundaries right — does each ticket's "must not" list match the neighbours you expect, and is nothing a ticket shouldn't do missing from it?
- Should any tickets be merged or split further?

Iterate until the user approves the breakdown.

### 5. Verify coverage before publishing

Before any ticket is published, check all three:

- **Requirement coverage** — every spec requirement (each user story, each acceptance criterion, each "must not") maps to at least one ticket, and every ticket records its **source requirements** (the spec sections it implements — see the template). Report anything uncovered to the user and get a decision, rather than silently dropping it.
- **Owner integrity** — every deferred item names a ticket that exists in this set and actually claims that work as its contract. A deferred item with no owner is a hole: either create the owning ticket or surface it to the user.
- **Invariant propagation** — every global invariant from the spec appears in every ticket that touches its code path.

Iterate until all three pass.

### 6. Publish the tickets to the configured tracker

Publish the approved tickets. **How** depends on the tracker `/setup-matt-pocock-skills` configured — the tickets are the same either way, only the shape of the blocking edges changes:

- **Local files** → write one file per ticket under `.scratch/<feature-slug>/issues/<NN>-<slug>.md`, numbered from `01` in dependency order (blockers first). Each file's "Blocked by" lists the numbers/titles it depends on. Use the per-ticket file template below — one ticket per file, never a single combined file.
- **A real issue tracker (GitHub, Linear, …)** → publish one issue per ticket in dependency order (blockers first) so each ticket's blocking edges can reference real identifiers. Use the platform's native blocking / sub-issue relationship where it has one; otherwise set each ticket's "Blocked by" to the blocking issues. Apply the `ready-for-agent` triage label unless instructed otherwise — the tickets are agent-grabbable by construction.

Work the **frontier**: any ticket whose blockers are all done. For local tickets, done means **Implementation status: completed** — all acceptance criteria checked, completion evidence complete, and a final review with no blocking findings. For a purely linear chain that means top to bottom.

Do NOT close or modify any parent issue.

<local-ticket-template>

# <NN> — <Ticket title>

**What to build:** the end-to-end behaviour this ticket makes work, from the user's perspective — not a layer-by-layer implementation list.

**Source requirements:** <the spec items this ticket implements, cited precisely — e.g. "User Stories 1, 3, 6, 47–49; Implementation Decisions / Runtime & API contracts bullets 2–7; Testing Decisions / integration tests bullets 1–5". Cite item and bullet numbers, not just section names — a reviewer must not have to guess the mapping>

**Blocked by:** the numbers/titles of the tickets that gate this one, or "None — can start immediately".

**Out of scope (deferred):** what this ticket must NOT do, naming the ticket that owns each item — especially any sibling ticket whose work looks adjacent to this one. "None" if the user approved none.

**Status:** ready-for-agent
**Implementation status:** pending

**Done means:** Implementation status flipped to **completed** — all acceptance criteria checked with complete completion evidence, /code-review returned no blocking findings, and the final review ran clean.

## Acceptance criteria (executable assertions)

- [ ] <behaviour that must work> — verified by <test at seam X / command Y>
- [ ] <negative: what must NOT happen> — verified by <the test asserting its absence>

**Completion evidence:** when checking a criterion, record the actual command run, its exit code, test counts, and any warnings. A criterion is not done if its verification was never run or could not be reproduced. New warnings, or warnings in ticket-owned tests, block completion; pre-existing unrelated warnings are recorded as baseline evidence and are not fixed unless the ticket scopes them.

**Global invariants (apply here):** <the spec's invariants this ticket's code path touches — e.g. no sensitive value in any output or log; no path escapes; stdout stays pure>. This ticket must satisfy them even where another ticket owns the full contract.

**Dependency/tooling guard:** <authorization, or "None — no dependency or tooling change is authorized by this ticket."> New direct dependencies, checkers, configs, or repo-level workflows not listed here are out of scope; any authorized dependency change must sync the repo's context/instruction file (AGENTS.md, CLAUDE.md, …), the lockfile, and relevant ADRs.

</local-ticket-template>

<issue-template>

## Parent

From showgp/skills-with-gate, MIT.

More from showgp/skills-with-gate