{"id":50412287,"url":"https://github.com/cskwork/supergoal-skill","last_synced_at":"2026-05-31T04:04:51.802Z","repository":{"id":361224704,"uuid":"1253597292","full_name":"cskwork/supergoal-skill","owner":"cskwork","description":"One objective in, a verified result out — a Claude Code skill that runs a full, gated dev process with expert subagents and refuses to declare done until a machine-checkable gate passes. Bilingual (EN/한국어) onboarding \u0026 live walkthrough: https://cskwork.github.io/supergoal-skill/","archived":false,"fork":false,"pushed_at":"2026-05-30T15:45:15.000Z","size":390,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-30T16:13:54.278Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"https://cskwork.github.io/supergoal-skill/","language":"JavaScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/cskwork.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-05-29T16:17:00.000Z","updated_at":"2026-05-30T14:26:50.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/cskwork/supergoal-skill","commit_stats":null,"previous_names":["cskwork/just-do-it-skill","cskwork/supergoal-skill"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/cskwork/supergoal-skill","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cskwork%2Fsupergoal-skill","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cskwork%2Fsupergoal-skill/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cskwork%2Fsupergoal-skill/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cskwork%2Fsupergoal-skill/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/cskwork","download_url":"https://codeload.github.com/cskwork/supergoal-skill/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/cskwork%2Fsupergoal-skill/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":33718492,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-05-31T02:00:06.040Z","response_time":95,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-05-31T04:04:51.283Z","updated_at":"2026-05-31T04:04:51.794Z","avatar_url":"https://github.com/cskwork.png","language":"JavaScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"# /supergoal\n\n**One objective in, a verified result out.**\nGive it a goal; it runs the full gated pipeline with expert subagents and refuses to declare success until a machine-checkable gate passes.\nNo extra install: clone the repo, symlink it into your skills directory, then `/supergoal \u003cobjective\u003e`.\nBest starting point: the **[landing page](https://cskwork.github.io/supergoal-skill/)** (bilingual English / 한국어, 3-step quickstart).\n\nA Claude Code skill that takes a single objective\nthrough a full, gated development process using expert subagents, then refuses to declare success\nuntil a machine-checkable gate passes.\n\nGated lanes, a single shared vault, an untrusted `claims.md` re-verified by an adversary, and a\nliteral-bash delivery gate that is never edited to pass. Each role's persona is a bundled file in\n`agents/`, so dispatch is **harness-agnostic**: it runs the same under Claude Code, Codex, agy, and\nother coding CLIs (the orchestrator spawns the persona via the harness's sub-agent mechanism, or runs\nit inline where none exists). **Nothing to install but the skill itself.** (Workflow inspired by\n[oh-my-symphony](https://github.com/cskwork/oh-my-symphony).)\n\n\u003e **New here? Start with the landing page** -\u003e **[cskwork.github.io/supergoal-skill](https://cskwork.github.io/supergoal-skill/)**\n\u003e A bilingual (English / 한국어) walkthrough with a 3-step quickstart, the three modes, how the\n\u003e builder-vs-verifier split catches real bugs, and the evidence it produces. Best onboarding path before you clone.\n\n## Modes\n\n`/supergoal` detects the mode from your objective:\n\n| Objective looks like | Mode | Pipeline |\n|---|---|---|\n| \"build / ship a new app/tool\" | **GREENFIELD** | Intake -\u003e **Validate (market/demand)** -\u003e Plan -\u003e **Human Feedback** -\u003e Build -\u003e Verify -\u003e QA -\u003e Deliver |\n| \"fix / broken / failing / why does\" | **DEBUG** | Intake -\u003e Reproduce -\u003e Diagnose -\u003e **Human Feedback** -\u003e Fix -\u003e Verify -\u003e Deliver |\n| \"add X to our existing/legacy code\" | **LEGACY** | Intake -\u003e Explore -\u003e Plan -\u003e **Human Feedback** -\u003e Build -\u003e Verify -\u003e QA -\u003e Deliver |\n| \"explain / understand / teach me X\" (learn, no code) | **LEARN** | Intake -\u003e Source -\u003e Bridge -\u003e Teach loop -\u003e Check (explain-back) -\u003e Journal |\n\n```text\n/supergoal build a habit-tracker app and ship it\n/supergoal the checkout page hangs intermittently in prod. fix it\n/supergoal add SSO to our legacy Django monolith\n```\n\n## Why it exists\n\nA single agent given a big objective drifts: it skips validation, trusts its own \"done\", and leaves\nunverified claims. `/supergoal` imposes the discipline a senior team would (see [`DESIGN.md`](DESIGN.md) and [`docs/research-brief.md`](docs/research-brief.md)):\n\n- **Topology, not preference, picks the architecture.** Fan out for wide-and-shallow work\n  (validation, scaffolding); single-driver for deep-and-narrow work (one bug, one feature).\n- **Builder != Verifier.** The agent that writes code never approves it. A fresh adversarial Verify\n  agent re-runs every `run-to-prove` from a clean state. (`claims.md` is untrusted.)\n- **Human Feedback before implementation.** After intake/repro/diagnosis/planning, the skill pauses\n  with two briefs: plain language first, then a novice-dev-friendly technical brief with term definitions.\n- **Two-layer done-gate.** Hard gate (tests/lint/build, deterministic) plus a soft committee\n  (architect + security + code-review). The rubric can never override a failing test.\n- **Gate on the project's own suite** (run in the workspace; the Verify agent independently re-runs from a clean state). Never benchmarks, never self-report.\n- **Bounded retry + circuit breaker.** Same error 3x trips the circuit breaker: stop, root-cause, escalate. No infinite loops.\n\n## The non-negotiable gates\n\n1. Validate-before-build (GREENFIELD).  2. Plan freezes scope.  3. Human Feedback approval.\n4. Builder != Verifier.  5. Multi-expert review before deliver.\n6. Literal delivery gate (`templates/delivery-gate.sh` exits 0).  7. Bounded retry + circuit breaker.\n\n## Install\n\nThis repo **is** the skill. Put it where Claude Code finds skills:\n\n```bash\ngit clone https://github.com/cskwork/supergoal-skill.git\n# then either symlink or copy it into your global skills dir:\nln -s \"$(pwd)/supergoal-skill\" ~/.claude/skills/supergoal\n# or: cp -R supergoal-skill ~/.claude/skills/supergoal\n```\n\nThen in Claude Code: `/supergoal \u003cyour objective\u003e`.\n\n## Layout\n\n```\nSKILL.md            thin spine: mode detection, gates, reference map\nagents/             one persona file per role (system prompt), harness-agnostic dispatch source of truth\nreference/          pipeline · experts · vault · market-research · quality-gates · debugging · qa · domain-rules · plan-grounding · learn\nreference/ui-ux.md  UI/UX overlay -\u003e taste-skill v2 (reference/taste-skill-v2.md, vendored) as design authority\nlearn/              LEARN-mode session journals (one file per session) + README template\ntemplates/          delivery-gate.sh · validate-gate.sh · human-feedback-gate.mjs · state.json\nDESIGN.md           research -\u003e decision mapping (cited)\ndocs/               research-brief.md · e2e-test-plan.md · changelog/ · index.html (landing)\nexamples/url-shortener/   a real service the harness built/debugged/extended (audit trail in docs/changelog/)\n```\n\n## Proof it works (live validation)\n\nAll three modes were run end-to-end on a real, production-grade service (a zero-dependency URL\nshortener, see [`examples/url-shortener/`](examples/url-shortener/), 68 tests). The audit trail for\neach run is in [`examples/url-shortener/docs/changelog/`](examples/url-shortener/docs/changelog/) (these early run records predate the file-set consolidation).\n\n- **GREENFIELD.** The adversarial Verify caught **2 real SSRF bypasses** (`[::ffff:127.0.0.1]`,\n  `localhost.`) and an unauth-500 that all passed the builder's own green tests, before shipping.\n- **DEBUG.** Given only a symptom (\"hits undercount under load\"), it reproduced (200 concurrent -\u003e\n  1/200), root-caused a **lost-update race**, stopped at Human Feedback for approval, fixed, and re-verified\n  with anti-flake concurrency runs (0 lost across 10 trials).\n- **LEGACY.** Added link-expiry (TTL) with **zero regressions** (backward-compatible with records\n  that predate the field), committee-approved, gate-green.\n\nAdversarial verification caught a real defect in 2 of 3 runs.\n\nA separate evidence-only private-codebase benchmark compared plain Codex CLI, `/supergoal`, and\nCodex Goal mode on the same hard backend task with the same hidden scorer. See\n[`docs/experiments/2026-05-30-private-codebase-comparison/`](docs/experiments/2026-05-30-private-codebase-comparison/).\n\n- **`/supergoal`:** passed all hidden checks, focused regressions, neighbor checks, `git diff --check`,\n  and the delivery gate.\n- **Codex Goal mode:** fixed the main code path and passed focused checks, but missed one hidden\n  fallback/preservation coverage check.\n- **Plain Codex CLI:** produced no usable result: idle run, no solution diff, no final output.\n\n## Credit\n\nConcept and workflow adapted from **oh-my-symphony** by cskwork\n(https://github.com/cskwork/oh-my-symphony). Built for Claude Code.\n\n## License\n\nMIT. See [`LICENSE`](LICENSE).\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcskwork%2Fsupergoal-skill","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fcskwork%2Fsupergoal-skill","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcskwork%2Fsupergoal-skill/lists"}