# Superpowers Issue Bridge > Connects a GitHub issue (created by brainstorm-to-issue) to Superpowers' brainstorming, writing-plans, and PR-creation stages, so the intent captured in the issue seeds the spec instead of being re-derived from scratch, and every downstream artifact traces back to the issue. Use whenever Superpowers' brainstorming or writing-plans is about to run — including when it is unknown whether an intent issue exists, since establishing that is this skill's first step — or when a PR is being opened for work that started from one. - Skill: `ismail9k/superpowers-issue-bridge` (Agent Skill) - Install (CLI): `npx skillmds@latest add ismail9k/superpowers-issue-bridge` - Raw SKILL.md: https://api.skillmd.com/api/skills/ismail9k/superpowers-issue-bridge/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: ismail9k (https://skillmd.com/u/ismail9k) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/ismail9k/superpowers-issue-bridge --- # Superpowers Issue Bridge ## Relationship to other skills — read this first - `brainstorm-to-issue` produces the intent issue: a short Problem / Proposed outcome / Affected users and systems / Constraints / Open questions write-up. It runs BEFORE any code-focused work starts. - Superpowers' own `brainstorming` skill produces the full `spec.md`: a longer interview covering implementation-relevant detail (alternatives considered, what's explicitly out of scope, edge cases) that intent capture doesn't attempt. - These are sequential, not competing. If the user says something like "let's brainstorm X" with no intent issue yet and no code context, that's `brainstorm-to-issue` territory. If an intent issue already exists and the user is ready to move to implementation planning, that's Superpowers' `brainstorming`, seeded by this bridge. When it's ambiguous which one applies, ask rather than guessing — don't let one silently absorb the other's job. ## Step 1 — Identify the intent issue Before invoking Superpowers' `brainstorming`, determine if an intent issue exists for this work: - The user references one directly ("build #42", "let's implement the claims-status issue"). - The current branch name encodes an issue number (e.g. `42-claims-status`). - Ask, if neither is present — don't assume there isn't one just because it wasn't mentioned, and don't guess a number. If there truly is no intent issue (e.g. a quick fix with no prior intent capture), proceed with Superpowers' `brainstorming` unmodified — this bridge has nothing to add. ## Step 2 — Seed brainstorming from the issue ```bash gh issue view --json title,body,url ``` Pass the issue's Problem / Proposed outcome / Affected users and systems / Constraints / Open questions into the `brainstorming` skill's context as already-answered. The interview should: - Not re-ask what the issue already states. - Still ask what the issue doesn't cover — alternatives, out-of-scope boundaries, technical constraints, edge cases. - Explicitly carry forward the issue's "Open questions" section as questions still needing an answer, unless the conversation resolves them. ## Step 3 — Trace the artifact back to the issue When `brainstorming` writes `spec.md`, add this to its header (alongside whatever front matter Superpowers already writes): ```markdown Intent-Issue: # — ``` When `writing-plans` writes `plan.md` from that spec, carry the same line forward unchanged. This is what makes `gh issue view ` and a repo search both lead back to the same thread later. ## Step 4 — Reference the issue in the PR When Superpowers' `finishing-a-development-branch` or `subagent-driven-development` is about to open the PR, ask the user one thing: **does this PR fully resolve the intent issue, or is it partial work toward it?** - Fully resolves it → include `Closes #` in the PR body (GitHub auto-closes the issue on merge). - Partial → include `Relates to #` instead (no auto-close; the intent issue stays open for further PRs). Don't default silently to `Closes` — an intent issue is often bigger than one PR, and auto-closing it prematurely breaks the audit trail this whole setup exists to protect. ## What NOT to do - Don't merge this bridge's logic into Superpowers' own skill files — keep it as a separate skill so Superpowers updates don't clobber it. - Don't let this skill re-interview the user on things the intent issue already answered. - Don't assume `Closes` vs `Relates to` — always ask once per PR. - Don't invent an intent issue number — if it can't be found or confirmed, proceed without one rather than guessing.