# replicate/cog — health index 98/100 (Exceptional)

> Inspection of the public repository replicate/cog by inspect.software. It holds a health index of 98 out of 100, placing it in the Exceptional band. The index is a signal derived from publicly visible practice, not a warranty, a security audit, or an endorsement, and it is independent of payment.

- Repository: https://github.com/replicate/cog
- Report page: https://inspect.software/software/replicate/cog
- Full JSON report: https://inspect.software/api/repositories/replicate/cog/report
- Badge: https://raw.githubusercontent.com/inspect-software/badges/main/v1/r/replicate/cog.svg
- Inspected: 2026-08-12
- Methodology: metrics 2.10.0, report schema 0.31.0 (https://inspect.software/methodology.md)

## Category summary

| Category | Weight | Index | Band |
|---|---|---|---|
| Vitality | 21% | 95/100 | Exceptional |
| Community & Adoption | 17% | 83/100 | Excellent |
| Sustainability & Governance | 23% | 92/100 | Excellent |
| Engineering Quality | 19% | 84/100 | Excellent |
| Security | 16% | 68/100 | Good |
| AI Readiness | 4% | 100/100 | Exceptional |

The weighted overall 86 is calibrated to 98 on the published index scale (record calibration 2026-08-02).

## Repository facts

| Field | Value |
|---|---|
| Description | Containers for machine learning |
| Classification | cli, library |
| Homepage | https://cog.run |
| Primary language | Go |
| License | Apache-2.0 |
| Stars | 9,455 |
| Forks | 695 |
| Created | 2021-02-26 |
| Last push | 2026-08-12 |
| Latest release | v0.21.0 |
| Commits (last year) | 400 |
| Bus factor | 4 |
| Topics | containers, cuda, docker, machine-learning, pytorch, tensorflow, ai |

## Published packages

| Registry | Package | Latest | Monthly downloads |
|---|---|---|---|
| go | github.com/replicate/cog | v0.21.0 | — |
| pypi | cog | 0.21.0 | — |

## Vitality — 95/100 (Exceptional)

Is the project alive — is code being written and are releases shipping? Weight: 21% of the overall index.

### Development activity — 91/100 (Excellent)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Push recency | met | 36 / 36 | last push 0 days ago |
| Commit cadence | partial | 27.7 / 36 | 40/52 weeks with commits |
| Commit volume | met | 18 / 18 | 400 commits in the last year |
| OpenSSF Scorecard: Maintained | excluded | 0 / 10 | OpenSSF Scorecard unavailable |

### Release discipline — 100/100 (Exceptional)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Ships releases | met | 27 / 27 | 100 releases published |
| Release recency | met | 36 / 36 | latest release 57 days ago |
| Release cadence | met | 27 / 27 | a release every ~6.9 days |
| OpenSSF Scorecard: Signed-Releases | excluded | 0 / 10 | OpenSSF Scorecard unavailable |

### Abandonment — 100/100 (Exceptional)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Project is still maintained | met | 100 / 100 | last human commit 10 days ago |

## Community & Adoption — 83/100 (Excellent)

Does the project have users, downloads, attention, and a welcoming setup for contributors? Weight: 17% of the overall index.

### Popularity & adoption — 94/100 (Exceptional)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Stars | met | 60 / 60 | 9,455 stars |
| Forks | partial | 23.7 / 25 | 695 forks |
| Watchers | partial | 10 / 15 | 65 watchers |

### Community health — 70/100 (Good)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| README | met | 22.5 / 22.5 | — |
| License | met | 22.5 / 22.5 | recognized license (Apache-2.0) |
| CONTRIBUTING guide | met | 18 / 18 | — |
| Code of conduct | not met | 0 / 13.5 | — |
| Issue template | not met | 0 / 7.2 | — |
| PR template | not met | 0 / 6.3 | — |

## Sustainability & Governance — 92/100 (Excellent)

Will the project survive its people — bus factor, responsiveness, who backs it, and package upkeep? Weight: 23% of the overall index.

### Maintainer resilience (bus factor) — 83/100 (Excellent)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Bus factor | partial | 43.2 / 54 | 4 contributor(s) cover half of all commits |
| Commit distribution | partial | 18.4 / 22.5 | top contributor authored 18% of commits |
| Contributor breadth | met | 13.5 / 13.5 | 82 contributors |
| OpenSSF Scorecard: Contributors | excluded | 0 / 10 | OpenSSF Scorecard unavailable |

### Issue & PR responsiveness — 88/100 (Excellent)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Issue resolution | partial | 40.4 / 42 | 96% of issues closed |
| PR acceptance | partial | 21.6 / 30 | 1618/2246 decided PRs merged |
| Newcomer PR acceptance | met | 13 / 13 | 1/1 first-time contributors' PRs merged in 30d |
| OpenSSF Scorecard: Code-Review | excluded | 0 / 15 | OpenSSF Scorecard unavailable |

### Ownership & stewardship — 100/100 (Exceptional)

Excluded from scoring (no data or not applicable): Verified domain. Remaining weights renormalized.

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Ownership backing | met | 30 / 30 | organization-owned |
| Verified domain | excluded | 0 / 20 | verified-domain status not read for this organization |
| Owner reach | met | 25 / 25 | 11,513 followers of replicate |
| Track record | met | 25 / 25 | 169 public repos, account ~6 yr old |

### Package maintenance — 100/100 (Exceptional)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Published & resolvable | met | 25 / 25 | 2 package(s) on go, pypi |
| Publish recency | met | 35 / 35 | latest publish 57 days ago |
| Version history | met | 20 / 20 | 248 published versions |
| Not deprecated | met | 20 / 20 | active, not deprecated or yanked |

## Engineering Quality — 84/100 (Excellent)

Are baseline engineering and documentation practices in place? Weight: 19% of the overall index.

### Engineering practices — 80/100 (Excellent)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| CI workflows | met | 24 / 24 | 11 workflow(s) |
| Tests present | met | 24 / 24 | — |
| Linter config | met | 16 / 16 | .golangci.yaml |
| Pre-commit hooks | not met | 0 / 9.6 | — |
| .editorconfig | not met | 0 / 6.4 | — |
| OpenSSF Scorecard: CI-Tests | excluded | 0 / 20 | OpenSSF Scorecard unavailable |

### Documentation — 90/100 (Excellent)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| README | met | 30 / 30 | — |
| Documentation directory | met | 25 / 25 | — |
| Documentation / homepage site | met | 15 / 15 | https://cog.run |
| Repository description | met | 10 / 10 | — |
| Topics | met | 10 / 10 | 7 topics |
| Wiki | not met | 0 / 10 | — |

## Security — 68/100 (Good)

Are visible security and supply-chain practices strong, with no malicious dependency and no unresolved high-risk jurisdiction exposure? Weight: 16% of the overall index.

### Security posture — 60/100 (Moderate)

Excluded from scoring (no data or not applicable): Dependency lockfiles. Remaining weights renormalized.

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Security policy (SECURITY.md) | not met | 0 / 30 | — |
| Dependabot config | met | 25 / 25 | — |
| Dependency lockfiles | excluded | 0 / 25 | published library — lockfiles are an application concern, not expected |
| CodeQL workflow | met | 20 / 20 | — |

### Dependency advisories — 100/100 (Exceptional)

Excluded from scoring (no data or not applicable): No advisories left outstanding. Remaining weights renormalized. Matched the pypi:cog@0.21.0 runtime dependency closure — what installing the published package pulls in — 9 packages. Reachability is not analyzed.

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Direct dependencies free of known advisories | met | 35 / 35 | no direct dependency carries a known advisory |
| Indirect dependencies free of known advisories | met | 25 / 25 | no indirect dependency carries a known advisory |
| No advisories left outstanding | excluded | 0 / 40 | no advisory carries a publication date |

### Malicious dependencies — 100/100 (Exceptional)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| No dependency reported as a malicious package | met | 100 / 100 | no dependency is reported as a malicious package |

### High-Risk Jurisdiction Exposure — 100/100 (Exceptional)

Only high-confidence self-published location evidence affects this multiplier. Ambiguous matches are review-only; country evidence is not proof of nationality, citizenship, legal registration, malicious intent, or sanctions status.

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Policy exposure multiplier | met | 100 / 100 | no confirmed policy-scope location match |

## AI Readiness — 100/100 (Exceptional)

How well is the repo equipped to be developed and maintained with AI coding agents? Carries a deliberately small weight: agent tooling is a real maintenance signal, but its absence must never gate the top of the scale (calibration saturates at raw 91, so 100/100 remains reachable with AI Readiness at zero).

### Agent context & guidance — 100/100 (Exceptional)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Agent instructions | met | 45 / 45 | AGENTS.md, CLAUDE.md |
| Machine-readable docs (llms.txt) | met | 15 / 15 | llms.txt present |
| Legible commit history | met | 40 / 40 | 83 of 84 human commits state their intent (structured subject or explanatory body) |

### Verify loop (build / test / typecheck) — 100/100 (Exceptional)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| One-command bootstrap | met | 18 / 18 | Makefile, mise.toml, noxfile.py |
| Automated tests | met | 22 / 22 | — |
| Lint / format config | met | 11 / 11 | .golangci.yaml |
| Static type checking | met | 11 / 11 | crates/coglet-python/coglet/py.typed, playground/tsconfig.json |
| Reproducible environment | met | 10 / 10 | lockfile |
| Demonstrated agent practice | met | 10 / 10 | 6 of the last 100 commits agent-authored or agent-credited |
| Automated maintenance | met | 8 / 8 | 15 of the last 100 commits are automated dependency updates |
| OpenSSF Scorecard: Pinned-Dependencies | excluded | 0 / 10 | OpenSSF Scorecard unavailable |

### Code legibility for models — 99/100 (Exceptional)

| Criterion | Status | Points | Detail |
|---|---|---|---|
| Type-checkable code | met | 45 / 45 | Go (statically typed) |
| Manageable file sizes | partial | 54.3 / 55 | 6/463 source files over 60KB |

### Machine-readable interfaces — 100/100 (Exceptional)

Excluded from scoring (no data or not applicable): API schema (OpenAPI/GraphQL/proto), MCP server. Remaining weights renormalized.

| Criterion | Status | Points | Detail |
|---|---|---|---|
| API schema (OpenAPI/GraphQL/proto) | excluded | 0 / 40 | not applicable to this kind of software |
| MCP server | excluded | 0 / 20 | not applicable to this kind of software |
| Runnable examples | met | 40 / 40 | examples, notebooks |

## Collection warnings

- Star history unavailable: GitHub GraphQL error: Resource not accessible by personal access token
- OpenSSF Scorecard did not return a usable result (exit code -9); skipping Scorecard checks

## What this report is

inspect.software measures public repositories against a single published, versioned methodology and reports the result as a 1–100 health index. Findings are signals derived from what is visible in public repository data — they are not a security audit, a warranty, or an endorsement, and no result depends on payment.

- Methodology, formulas and weights: https://inspect.software/methodology.md
- What a result does and does not claim: https://inspect.software/wiki/signals-not-warranties.md
- Rating bands: https://inspect.software/wiki/scoring-bands.md
- Corrections: https://inspect.software/contact.md
- Index of all documents: https://inspect.software/llms.txt