Skip to content

docs(scorecard): drop Branch-Protection PAT recommendation #274

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
blove merged 2 commits into from
Jun 25, 2026
Merged

docs(scorecard): drop Branch-Protection PAT recommendation #274

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7389834
docs(scorecard): drop Branch-Protection PAT recommendation
blove Jun 25, 2026
efcb194
docs(scorecard): align current-board action cell with no-PAT stance
blove Jun 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion docs/superpowers/plans/2026-06-18-ossf-scorecard-uplift.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -551,7 +551,7 @@ After merge to `main`, confirm Scorecard's Vulnerabilities check rises from 0 to
## Out of scope (tracked elsewhere — see spec)

These need Brian's GitHub account and are **not** implemented by this plan:
- **Branch-Protection** (-1): create a fine-grained PAT with admin:read, store as the `repo_token`/`SCORECARD_TOKEN` secret, pass it to `ossf/scorecard-action`.
- **Branch-Protection** (-1): **leave inconclusive — do NOT wire a PAT.** A `-1` is excluded from the Scorecard aggregate mean, so it costs nothing. A *scored* check for this repo's intentionally-weak protection (solo maintainer, `gh pr merge --auto --squash`, `strict: false`, no required reviews, `enforce_admins: false`) would land ~4 and drag the aggregate down. A standing `SCORECARD_TOKEN`/`repo_token` admin:read PAT is also CI supply-chain surface. AAF (`cacheplane/angular-agent-framework`) wired it (#689) then reverted (#708) for exactly this reason — keep `scorecard.yml` token-less.
- **Code-Review** (0): route future `main` commits through reviewed PRs (start by reviewing/merging Phase 1 and Phase 2 as PRs).
- **CII-Best-Practices** (0): self-certify at bestpractices.dev.
- **Maintained** (0): auto-resolves ~2026-07-14 (90-day repo age) — no action.
Expand Down
10 changes: 6 additions & 4 deletions docs/superpowers/specs/2026-06-18-ossf-scorecard-uplift-design.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ Scorecard is a weighted average of per-check scores (Critical=10, High=7.5, Medi
| Pinned-Dependencies | 0 | Medium | GitHub Actions pinned to tags (`@v6`), not commit SHAs | PR 1 |
| SAST | 0 | Medium | No CodeQL (or other recognized SAST) workflow | PR 1 |
| Vulnerabilities | 0 | High | 25 OSV advisories in dev/docs deps (Next.js ×8, vitest, vite, esbuild, turbo, ws, langsmith, js-yaml, uuid) | PR 2 |
| Branch-Protection | -1 | High | Default `GITHUB_TOKEN` can't read classic branch-protection rules | Follow-up (PAT) |
| Branch-Protection | -1 | High | Default `GITHUB_TOKEN` can't read classic branch-protection rules | None — leave `-1` (excluded); do NOT wire a PAT (see follow-up #1) |
| Code-Review | 0 | High | 0/23 recent changesets went through an approved PR | Follow-up (process) |
| CII-Best-Practices | 0 | Low | No OpenSSF Best Practices badge | Follow-up (badge) |
| Maintained | 0 | High | Repo created <90 days ago (2026-04-14) | **Auto-resolves ~2026-07-14** |
Expand Down Expand Up @@ -56,7 +56,7 @@ Expected: Vulnerabilities 0→10. ≈ +0.86 aggregate. The check stays near 0 un

These are split off because they need account-level actions an agent can't perform:

1. **Branch-Protection (PAT).** Create a fine-grained PAT with **read** access to repository administration, store as the `repo_token` (or `SCORECARD_TOKEN`) secret, and pass it to `ossf/scorecard-action`. Un-excludes Branch-Protection (-1 → ~8, High weight, ≈ +0.85). The repo already has required reviews + status checks + conversation resolution; strengthening with `enforce_admins` and required signatures would push the tier higher.
1. **Branch-Protection — leave inconclusive (`-1`); do NOT wire a PAT.** It is tempting to create a fine-grained PAT (`Administration: read`) stored as `SCORECARD_TOKEN`/`repo_token` so `ossf/scorecard-action` can read branch-protection settings (the default `GITHUB_TOKEN` can't, which is *why* the check is `-1`). **Don't.** An inconclusive `-1` is **excluded from the Scorecard aggregate mean**, so it costs nothing. Dawn is a solo-maintainer repo that merges via `gh pr merge --auto --squash` with intentionally weak protection (`strict: false`, `required_pull_request_reviews: null`, `enforce_admins: false`); a *scored* Branch-Protection check for that posture lands ~4 (a Medium tier) and would **drag the aggregate down** rather than up. A standing admin:read PAT is also CI supply-chain surface. The sibling repo `cacheplane/angular-agent-framework` proved this empirically: it wired the PAT in #689, then reverted it in #708 ("drop SCORECARD_TOKEN — let Branch-Protection go inconclusive"). Net: keep `scorecard.yml` token-less and let Branch-Protection stay `-1`.
2. **Code-Review (process).** Route your own commits to `main` through reviewed PRs going forward so approved changesets accumulate. PR 1 and PR 2 themselves should be merged this way to start the count. Climbs 0 → up over time (High weight).
3. **CII-Best-Practices (badge).** Self-certify at [bestpractices.dev](https://bestpractices.dev) (free). 0 → 5+ (Low weight).

Expand All @@ -74,8 +74,10 @@ These are split off because they need account-level actions an agent can't perfo
| Today | 4.1 |
| After PR 1 (config) | ~6.2 |
| After PR 2 (deps) | ~7.0 |
| After follow-ups (PAT + Code-Review + badge) | ~7.8 |
| ~2026-07-14 (Maintained auto-flips) | ~8.6 |
| After follow-ups (Code-Review + badge) | ~7.0 |
| ~2026-07-14 (Maintained auto-flips) | ~7.8 |

Branch-Protection is intentionally **excluded** from this trajectory: it stays inconclusive (`-1`, dropped from the aggregate mean) rather than being un-excluded by a PAT — see follow-up #1. The earlier ≈ +0.85 "un-excludes Branch-Protection" estimate was wrong-signed: a *scored* check for this repo's weak protection would lower the mean, not raise it.

## Verification

Expand Down