oliverspec-scope
Skill0 starsChanged 3 months agoMITView on GitHub
Scope an OliverSpec PRD + TRD into Linear tickets with clear descriptions, acceptance criteria, and explicit blockedBy/blocks relations. Use after oliverspec-propose when the user wants to scope, /oliverspec scope, or oliverspec-scope. Does not write proposal docs to the git repo.
In this repository9 other files it gives its agents.
- babysitSkill
- dashboard-uiSkill
- factorySkill
- lab-notebookSkill
- oliverspec-applySkill
- oliverspec-babysitSkill
- oliverspec-exploreSkill
- oliverspec-proposeSkill
- oliverspec-shipSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 5 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
When scoped tickets are ready, run oliverspec-apply.
- Distinctive: line 45
Read Linear MCP tool schemas before calling tools
- Very specific: line 48
If either document is missing or too thin to slice, stop — send the user back to oliverspec-propose
- Distinctive: line 64
Every ticket gets testable acceptance criteria
- Distinctive: line 68
Prefer TRD section boundaries as natural ticket seams when they map cleanly
- Very specific: line 112
Critical: After create, every dependency edge must be applied with save_issue blockedBy and/or blocks.
- Very specific: line 116
Verify each planned edge (get_issue with includeRelations: true if needed)
- Distinctive: line 119
Incomplete without relations: If the Mermaid graph or description implies edges missing from Linear relations, this skill is not done — fix …
- Distinctive: line 148
Do not start implementation in this skill
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Scope an OliverSpec PRD + TRD into Linear tickets. Never write proposal markdown into the git repo.
Prerequisite: a Linear project with PRD and TRD documents (from oliverspec-propose). If missing, stop and tell the user to run oliverspec-propose first.
When scoped tickets are ready, run oliverspec-apply.
Defaults
| Setting | Default |
| Linear team | Engineering (override only if user specifies another team) |
| Source of truth | Project PRD + TRD Linear documents |
| PR stacking later | GitHub stacked PRs (gh stack) — slice tickets to be stack-friendly |
Input
- Linear project name (or infer from conversation)
- Optional: constraints on ticket size / sequencing
Announce: Scoping OliverSpec project: <name>.
Steps
1. Load PRD + TRD
- Read Linear MCP tool schemas before calling tools
- Resolve the project (
list_projects/get_project) list_documents+get_documentfor the PRD and TRD- If either document is missing or too thin to slice, stop — send the user back to
oliverspec-propose
2. Draft the scoped ticket breakdown (in chat first)
From the PRD (product what/why) and TRD (deeply technical how), draft and show:
- Numbered ticket list — title, scope summary, ordering rationale (cite PRD/TRD sections)
- Mermaid dependency graph of tickets (flowchart — never ASCII)
- Acceptance criteria preview per ticket (checkboxes)
Principles for slicing:
- Each ticket is independently reviewable as its own stacked PR
- Order by dependency — schema before logic, backend before frontend
- "Flip the switch" last — enforcement/gating after supporting flow
- Parallel where possible — call out tickets that don't block each other
- Every ticket gets testable acceptance criteria
- TDD by default — tickets are implemented with test-driven development in
oliverspec-apply. Scope each ticket so acceptance criteria demand comprehensive tests for a full implementation (happy path, edge cases, failure modes called out in the TRD). Thin “add a smoke test” ACs are not enough - Stack-friendly slices —
oliverspec-applyopens one stacked PR per ticket; avoid grab-bag tickets that fight rebases - Prefer chains over diamonds — a GitHub stack is a strictly linear chain, so a ticket with two independent
blockedByparents can only sit under one of them. Wire relations as chains where the real dependency allows it, and only use a second parent when the ticket genuinely needs both - Prefer TRD section boundaries as natural ticket seams when they map cleanly
Ask the user to confirm before creating anything in Linear.
3. Create tickets
For each confirmed ticket, save_issue:
team: Engineering (or override)project: the project name/idtitle: concise, implementation-orienteddescription: use the ticket template belowpriority: set when obvious; otherwise omit
Ticket template
## Context
<Why this ticket exists; link PRD + TRD themes/sections>
## Scope
- <concrete sub-deliverable>
- <...>
## Acceptance Criteria
- [ ] <observable, testable product/behavior outcome>
- [ ] <...>
- [ ] Tests first (TDD): comprehensive failing tests for the behavior above
- [ ] Typecheck stays green via stubs/fakes/interfaces while those tests are still red
- [ ] Full implementation replaces stubs until tests pass
- [ ] Relevant typecheck/lint/test commands pass for this ticket’s surface
## Dependencies
<!-- Documentary only — you MUST also set Linear blockedBy/blocks relations -->
- Blocked by: <ticket titles or ids>
- Blocks: <...>
## Notes
<edge cases, out of scope for this ticket, implementation hints from TRD>Include concrete test expectations in ACs when the TRD names them (e.g. module, scenario, regression). Don’t leave testing as an optional afterthought.
4. Wire Linear relations (required)
Critical: After create, every dependency edge must be applied with save_issue blockedBy and/or blocks. Mentioning blockers only in the description is not enough.
- Set
blockedBy/blocksviasave_issueupdates using issue identifiers - Description
## Dependenciesis documentation only; native relations are the source of truth for apply waves - Verify each planned edge (
get_issuewithincludeRelations: trueif needed) - Optionally update the PRD/TRD with a Ticket index table (IDs, titles, depends-on, links)
Incomplete without relations: If the Mermaid graph or description implies edges missing from Linear relations, this skill is not done — fix relations before handoff.
5. Move project to Planned
After tickets are created (and relations wired), set the Linear project status to Planned (save_project with id = the project and state: "Planned", or the team's equivalent planned-stage state if the name differs). Do not move it earlier — Backlog/idea status stays until the ticket set actually exists. This scope pass is what turns the project from an idea into a plan.
6. Done — hand off
Summarize:
- Project + PRD + TRD links
- Project status: Planned
- Ticket table: ID, title, blocked by, blocks, URL
- Prompt: "Run
oliverspec-apply(optionally with the project name) to implement — parallel sub-agents and one stacked PR per ticket."
Guardrails
- Read PRD + TRD first — don't invent scope that isn't grounded in those docs
- Move the project to Planned after tickets are created (
save_projectstate) — not before; don't leave it in Backlog once the ticket set exists - No repo proposal artifacts
- Confirm breakdown before creating issues
- Default team Engineering
- Diagrams in Mermaid only
- Every ticket must have acceptance criteria as checkboxes
- TDD-ready tickets — ACs must require comprehensive tests for a full implementation; apply will implement via TDD
- Explicit Linear blockers required —
blockedBy/blockson every dependency edge - Do not rewrite the PRD/TRD except optional ticket index / links — design changes go back to
oliverspec-propose - Do not start implementation in this skill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

