|
| 1 | +# OpenSSF Security Baseline |
| 2 | + |
| 3 | +This section records how Panurus measures up against the |
| 4 | +[OpenSSF Open Source Project Security Baseline](https://baseline.openssf.org) (OSPS Baseline), so that |
| 5 | +contributors and consumers can see which security practices the project already follows, which ones |
| 6 | +are still missing, and how to re-run the assessment. |
| 7 | + |
| 8 | +## Two different OpenSSF programs |
| 9 | + |
| 10 | +The two OpenSSF programs that apply to a project like Panurus are often confused. They are separate, |
| 11 | +and only the first one is assessed here. |
| 12 | + |
| 13 | +| Program | Identifiers | Levels | Used here | |
| 14 | +|---------|-------------|--------|-----------| |
| 15 | +| [OSPS Baseline](https://baseline.openssf.org) | `OSPS-<CATEGORY>-<NN>.<NN>` | Level 1, Level 2, Level 3 | Yes — the pages below | |
| 16 | +| [Best Practices Badge](https://www.bestpractices.dev) | free-form criteria | passing, silver, gold | No — tracked separately in the project's [badge entry](https://www.bestpractices.dev/en/projects/7176) | |
| 17 | + |
| 18 | +The OSPS Baseline levels are *not* named "passing", "silver" or "gold" — those are Best Practices |
| 19 | +Badge tiers. Baseline levels are scoped by project size instead: |
| 20 | + |
| 21 | +- **Level 1** — any code or non-code project, any number of maintainers or users. |
| 22 | +- **Level 2** — a code project with at least two maintainers and a small, consistent user base. |
| 23 | +- **Level 3** — a code project with a large, consistent user base. |
| 24 | + |
| 25 | +Every Baseline control is a `MUST`; the Baseline deliberately contains no `SHOULD` entries. |
| 26 | + |
| 27 | +## What Panurus targets |
| 28 | + |
| 29 | +Panurus has several active maintainers (see [MAINTAINERS.md](../../MAINTAINERS.md)), tagged releases, |
| 30 | +and downstream users, so **Level 2 is the level the project aims to satisfy in full**. Level 1 is |
| 31 | +almost entirely satisfied today; Level 3 is documented as a longer-term target because it requires |
| 32 | +release-signing, SBOM, VEX and threat-modeling work that has not started. |
| 33 | + |
| 34 | +## Assessment |
| 35 | + |
| 36 | +- **Assessed against:** OSPS Baseline **v2026.02.19** (the current release at the time of writing) |
| 37 | +- **Assessment date:** 2026-08-03 |
| 38 | +- **Assessed release:** `v0.16.0` |
| 39 | + |
| 40 | +| Level | Controls | Met | Partially met | Not met | Unverified | |
| 41 | +|-------|---------:|----:|--------------:|--------:|-----------:| |
| 42 | +| [Level 1](baseline_level_1.md) | 24 | 20 | 2 | 0 | 2 | |
| 43 | +| [Level 2](baseline_level_2.md) | 19 | 11 | 5 | 2 | 1 | |
| 44 | +| [Level 3](baseline_level_3.md) | 21 | 3 | 6 | 12 | 0 | |
| 45 | + |
| 46 | +Status values used in the per-level tables: |
| 47 | + |
| 48 | +| Status | Meaning | |
| 49 | +|--------|---------| |
| 50 | +| **Met** | Satisfied, with evidence in this repository or in a publicly verifiable GitHub setting. | |
| 51 | +| **Partially Met** | Partly satisfied; the remaining gap is named in the notes. | |
| 52 | +| **Not Met** | Not satisfied today. | |
| 53 | +| **Unverified** | Cannot be confirmed from public repository state; needs confirmation by a maintainer or org administrator. | |
| 54 | + |
| 55 | +Nothing is marked **Met** on the basis of "this is standard practice for the foundation". A control is |
| 56 | +only **Met** when a file in this repository, a release artifact, or a publicly readable GitHub |
| 57 | +setting shows it. |
| 58 | + |
| 59 | +## Main gaps |
| 60 | + |
| 61 | +The assessment converges on a small number of themes rather than 20 unrelated items: |
| 62 | + |
| 63 | +1. **Release provenance.** Releases carry no signed manifest, no checksums and no SBOM, and the tags |
| 64 | + are lightweight rather than signed — so there is nothing for a consumer to verify, and no |
| 65 | + documented verification procedure (`OSPS-BR-06.01`, `OSPS-DO-03.01`, `OSPS-DO-03.02`, |
| 66 | + `OSPS-QA-02.02`). |
| 67 | +2. **Support lifecycle.** No statement of how long a release is supported or when it stops receiving |
| 68 | + security fixes (`OSPS-DO-04.01`, `OSPS-DO-05.01`). |
| 69 | +3. **Security analysis artifacts.** No published security assessment or threat model, and the |
| 70 | + project README states it has not been audited (`OSPS-SA-03.01`, `OSPS-SA-03.02`). |
| 71 | +4. **Policy thresholds.** CodeQL, `golangci-lint` and Dependabot all run, but no document defines the |
| 72 | + severity threshold at which findings must be fixed, or what happens before a release |
| 73 | + (`OSPS-VM-05.01`, `OSPS-VM-05.02`, `OSPS-VM-06.01`). |
| 74 | +5. **Enforcement vs. policy.** [DEVELOPMENT.md](../../DEVELOPMENT.md) requires one maintainer approval |
| 75 | + per PR, but the active branch ruleset requires zero approving reviews and only the DCO check, so |
| 76 | + the policy is honored by convention rather than enforced by the platform (`OSPS-QA-03.01`, |
| 77 | + `OSPS-QA-07.01`). |
| 78 | +6. **Undeclared workflow permissions.** Most workflows declare least-privilege `permissions:`, but |
| 79 | + `tests.yml`, `md_links.yml` and `protect-integration-test-types.yml` do not, and one |
| 80 | + `workflow_dispatch` input is interpolated straight into a shell step (`OSPS-AC-04.01`, |
| 81 | + `OSPS-AC-04.02`, `OSPS-BR-01.04`). |
| 82 | + |
| 83 | +## How to reassess |
| 84 | + |
| 85 | +1. Check whether a newer Baseline release exists at |
| 86 | + [baseline.openssf.org](https://baseline.openssf.org). Only the version labeled *current* should |
| 87 | + be used for new compliance work, and control identifiers are occasionally retired between |
| 88 | + versions. |
| 89 | +2. Pull the machine-readable checklist for that version (for example |
| 90 | + `https://baseline.openssf.org/versions/2026-02-19-checklist.md`) and diff its control list against |
| 91 | + the tables in [Level 1](baseline_level_1.md), [Level 2](baseline_level_2.md) and |
| 92 | + [Level 3](baseline_level_3.md). |
| 93 | +3. Re-verify each row against the repository, not against memory. Controls about GitHub |
| 94 | + configuration can be checked with the API, for example: |
| 95 | + |
| 96 | + ```bash |
| 97 | + # branch rulesets: which checks and approvals are actually enforced on main |
| 98 | + gh api repos/LFDT-Panurus/panurus/rulesets |
| 99 | + gh api repos/LFDT-Panurus/panurus/rulesets/<id> |
| 100 | + |
| 101 | + # release assets, signatures and changelog |
| 102 | + gh release view v0.16.0 --json tagName,assets,body |
| 103 | + |
| 104 | + # private vulnerability reporting and published advisories |
| 105 | + gh api repos/LFDT-Panurus/panurus/private-vulnerability-reporting |
| 106 | + gh api repos/LFDT-Panurus/panurus/security-advisories |
| 107 | + ``` |
| 108 | + |
| 109 | +4. Update the header of each page (Baseline version, assessment date, assessed release) and the |
| 110 | + summary table above. |
| 111 | +5. Open a GitHub issue for every control that moves to, or stays at, **Not Met**, and reference the |
| 112 | + control identifier in the issue so progress stays traceable. |
| 113 | +6. If the project also wants credit on the Best Practices Badge, update |
| 114 | + [project 7176](https://www.bestpractices.dev/en/projects/7176) separately — the two programs do |
| 115 | + not share data. |
| 116 | + |
| 117 | +## Related documentation |
| 118 | + |
| 119 | +- [Security policy](../../SECURITY.md) — how to report a vulnerability |
| 120 | +- [Contributing](../../CONTRIBUTING.md) and [Development guidelines](../development/general.md) |
| 121 | +- [Selector resource limits](../security/selector_resource_limits.md) — a security-relevant design note |
0 commit comments