The capstone. Reads every audit report in the repo, dedupes overlapping findings, and merges them into one sequenced plan.
Free & open · no signup · read-only — it ends by asking · nothing leaves your machine
You are working inside this repo. Mission: read every audit report at the repo root and in reports/ — the outputs of prior goal prompts — and merge them into one deduplicated, dependency-aware, sequenced roadmap.
This is synthesis, not re-auditing: work from the reports, dipping into code only to resolve conflicts. Your only write is the report file.
One structured report at the repo root — or in reports/, if you keep one — the same shape every time, ready for a teammate — or the optional Studio — to act on.
Every Goal Prompt follows the same four steps, so results are consistent and repeatable — no matter which one you run.
Find the audit reports at the repo root and in reports/: IMPROVEMENTS.md, BUGS.md, PERF.md, SECURITY-AUDIT.md, TESTING…
Extract every item into one pool.
Score globally.
Create ROADMAP.md at repo root.
Copy it and paste it into your agent inside the repo you want checked.
Install the goal plugin once — two commands — then just type /goal:roadmap-synthesis.
/plugin marketplace add GhostlyGawd/goal-prompts/plugin install goal@goal-promptsOr install only this Goal Prompt as /goal-roadmap-synthesis:
curl -fsSL https://goal-prompts.vercel.app/install | BRIEF=28 shLet an agent fetch it mid-conversation, or pull the raw Goal Prompt by URL.
https://goal-prompts.vercel.app/raw/28.mdNothing hidden — this is the whole Goal Prompt, verbatim. Read it in a minute, edit it, or copy it as-is.
# Goal: Roadmap Synthesis You are working inside this repo. Mission: read every audit report at the repo root and in `reports/` — the outputs of prior goal prompts — and merge them into one deduplicated, dependency-aware, sequenced roadmap. This is synthesis, not re-auditing: work from the reports, dipping into code only to resolve conflicts. Your only write is the report file. ## Phase 1 — Collect - Find the audit reports at the repo root and in `reports/`: IMPROVEMENTS.md, BUGS.md, PERF.md, SECURITY-AUDIT.md, TESTING.md, and any others matching this family. - List what exists and what's missing; note which absent audits would most change this roadmap. ## Phase 2 — Normalize - Extract every item into one pool: source report · name · impact · effort · tags. - Dedupe: the same underlying issue often surfaces in multiple audits (a god file flagged by DEBT and HOTSPOTS; a slow query in PERF and COSTS). Merge them, keeping all evidence and the strongest framing. - Find themes: many items, one root cause — name the root cause as its own item. ## Phase 3 — Sequence - Score globally: impact, effort, risk — recalibrated across all reports on one scale. - Map dependencies BETWEEN items: fixes that unlock others, refactors that make features cheap, tests that de-risk migrations. - Balance the mix: pure risk-reduction and pure growth are both losing strategies; interleave them. ## Phase 4 — Report Create `ROADMAP.md` at repo root: 1. **Sources** — reports found, reports missing, and the audit to run next 2. **Unified backlog** — deduped table: item · sources · impact · effort · dependencies 3. **Themes** — root causes worth one structural fix instead of five patches 4. **Three milestones** — Now (1–2 weeks) · Next (a month) · Later — each with its items and the one-line story of why this order 5. **Merge log** — what was deduplicated, so nothing looks lost Start the report with today's date. If `ROADMAP.md` already exists from a previous run, read it first and lead with what changed since. ## Rules - Every roadmap item traces to a source report — no new findings smuggled in - When reports disagree on priority, say so and rule with reasoning - If a `reports/` directory exists at the repo root, write the report there instead of the root. - Before asking, present the top findings as a ranked list in plain words - Report only — end by asking whether to adjust the sequence
reports/ directory exists at the repo root, write the report there instead of the root.Curated neighbors — briefs that answer the adjacent question, worth running in the same session.
Runs inside these playbooks — curated sequences you can launch with one paste: