Is Claude Agent SDK safe to use?

Claude Agent SDK has an HVTrust score of 72.4/100 (Grade B), ranked #300 of 1371 tracked AI agent projects. This score reflects verifiable supply-chain integrity, transparency, maintenance, and adoption signals.

What is the trust score for Claude Agent SDK?

Claude Agent SDK scores 72.4/100 on HVTracker's evidence-weighted HVTrust scale. It has 8.2k GitHub stars and is categorized under Agent Frameworks.

Claude Agent SDK

Repository by anthropics

Agent Frameworks Python Listed MIT

Is Claude Agent SDK safe? Promising trust profile, but some evidence still deserves review.

OSSF Scorecard 5.0 / 10

Provenance None

Signed commits 16%

Last push 2d ago

Advisories None known

Compare Claude Agent SDK

How does it stack up against its Agent Frameworks neighbours?

Pick any agent to compare →

Suggest a correction

In detail: Claude Agent SDK scores 72.4/100 (Grade B), ranked #300 of 1371 tracked open-source AI agent projects, on evidence coverage B (3 of 5 independent signal types). The public evidence: no package-provenance attestation found; OSSF Scorecard rates its supply-chain practices 5.0/10; 16% of recent commits are signed; no published advisory affects its latest release (claude-agent-sdk); last pushed 2026-10-09. Every point is earned from checkable signals — never paid placement. How scoring works →

How Claude Agent SDK could raise its score

Each line changes one public signal and recomputes with the live scoring function. The gains don't add up exactly, because the score is capped near the top.

Publish build provenance for its packagesPyPI Trusted Publishing attestations +13.7 → 86.1 A · #10 of 129
Raise the OSSF Scorecard from 5.0 to 9.0lowest checks: CII-Best-Practices 0, Dependency-Update-Tool 0, Fuzzing 0 +8.2 → 80.6 A · #14 of 129
Sign every commit16% signed today +4.1 → 76.5 B · #17 of 129

Already done? Suggest a correction Add the trust badge

Chain of custody

Scanners check what the code says. This traces who ships Claude Agent SDK and whether that has changed: from the source repository, through how changes are reviewed and released, to the package you install. These are the checks for OWASP ASI04 Agentic Supply Chain Vulnerabilities.

Source Partly signed

github.com/anthropics/claude-agent-sdk-python, MIT, last pushed 2026-10-09. 16% of the last 100 commits carry a verified signature; the rest can't be tied to a verified identity.
Review and release Mixed

OpenSSF Scorecard rates the repository's practices 5.0/10 (scanned Oct 9, 2026). The checks that decide who can get a change released:
- Code Review2
- Branch Protection6
- Signed Releasesn/a
- Dangerous Workflow10
- Token Permissions0
- Pinned Dependencies1
All 18 Scorecard checks

Binary-Artifacts 10

Branch-Protection 6

CI-Tests 10

CII-Best-Practices 0

Code-Review 2

Contributors 10

Dangerous-Workflow 10

Dependency-Update-Tool 0

Fuzzing 0

License 10

Maintained 10

Packaging -1

Pinned-Dependencies 1

SAST 0

Security-Policy 0

Signed-Releases -1

Token-Permissions 0

Vulnerabilities 10
Published packages Linked, not attested
- PyPI claude-agent-sdk Source link points back to this repo No build attestation
Without a build attestation, nothing proves the published files were built from this repository.
What you install today No known advisories

Latest release checked: claude-agent-sdk 0.2.165. No published advisory affects it (OSV, checked 2026-10-11).

Changes to this chain

No change to package provenance, package source links, Scorecard coverage or license in HVTracker's daily snapshots of Claude Agent SDK.

Evidence behind this score

Coverage B: 3 of 5 independent evidence types found.

GitHub repository data
Package downloads
Supply-chain checks
Public actions (missing)
Community mentions (missing)

Verify this score yourself

Every build re-issues Claude Agent SDK’s score as an Ed25519-signed credential, valid for 7 days. You can check it offline against HVTracker’s published key; if anyone changes a number after signing, verification fails.

Signed credential (JSON) How to verify

What this score doesn’t check

It covers who ships the code, not what the code does. It doesn’t read tool descriptions or prompts for injected instructions, watch runtime behaviour, or find bugs nobody has disclosed yet. For that, run a content scanner before you connect it, such as Cisco MCP Scanner or Snyk Agent Scan.

How Claude Agent SDK compares in Agent Frameworks

Bars show each HVTrust score; the tick marks Claude Agent SDK’s 72.4.

Where the 72.4 comes from

HVTrust dimensions vs the Agent Frameworks average

72.4 / 100 · 100.0% confidence

Claude Agent SDK Agent Frameworks average (129 agents)

Safety / Integrity50% OSSF Scorecard · 30% provenance · 20% signed commits

7.1 / 25

2.6 below avg 9.7

Identity / Provenance60% listing status · 40% build provenance

10.8 / 18

0.8 below avg 11.6

Transparency50% declared license · 50% OSSF Scorecard

12.8 / 17

in line with avg 12.3

Maintenance60% last-push freshness · 40% commit activity

19.2 / 20

4.9 above avg 14.3

AdoptionLog-scaled stars · package downloads

18.6 / 20

7.4 above avg 11.2

Quick Trust Read

What Would Improve It

Publish package provenance or release attestations for stronger supply-chain evidence.

Recent Changes

2026-10-08

Rank Moved

Rank rose 13 spots (#300 → #287)

2026-09-23

Rank Moved

Rank dropped 177 spots (#113 → #290)

2026-09-21

Rank Moved

Rank rose 177 spots (#290 → #113)

Maintainer Checklist

Raise Scorecard signals Current OSSF Scorecard is 5.0/10. Tighten the weakest checks to improve public safety evidence.

Publish provenance Add package provenance or release attestations so users can verify where shipped artifacts came from.

Increase signed commits Raise the share of verified-signed commits to make maintainer identity and release history easier to trust.

Compare with another agent Get email alerts for Claude Agent SDK View Agent Frameworks Report missing data

85.8

Activity sub-score · out of 100

#20

In Agent Frameworks

How to read this: HVTrust (0–100) weighs supply-chain signals (provenance, OSSF Scorecard, signed commits, open license) alongside real-world adoption. Grade B reflects the trust score band: A ≥ 80, B ≥ 65, C ≥ 50, D < 50. Evidence coverage B is separate — it grades how many independent signal types back the score (3 of 5), so a high score on thin evidence stays visible. Full methodology →

Rank Trend

2026-09-29 2026-10-11

Activity & Reach

Stars

8.2k

Forks

1.3k

Last Push

2026-10-09

2 days ago

Commits (4 wk)

Downloads (7d)

7,639,787

pypi

HN mentions (30d)

—

Open Issues

216

Rank Change

▼4

was #296

Analysis

Activity Inputs

85.8 / 100

StarsRepository reach

23.5 / 30

FreshnessLast push recency

24.7 / 25

ActivityRecent commits

22.9 / 25

CommunityFork signal

14.5 / 20

Common questions about Claude Agent SDK

Claude Agent SDK has a mixed signal profile. Some trust indicators are present, others are missing. Whether it is safe for your use case depends on which gaps matter to you — review the breakdown below before adopting in production.

Does Claude Agent SDK publish package provenance?

No published build provenance is currently detected for Claude Agent SDK. This is common for open-source projects but means consumers cannot independently verify that the package on the registry matches the GitHub source.

Does Claude Agent SDK have an OpenSSF Scorecard?

Claude Agent SDK has an OpenSSF Scorecard score of 5.0/10. The Scorecard checks for branch protection, signed releases, dependency updates, fuzzing, code review, and other supply-chain hygiene items. See the full check breakdown on this page.

Is Claude Agent SDK actively maintained?

Actively maintained. The repository was pushed to within the last 2 day(s).

What license does Claude Agent SDK use?

Claude Agent SDK ships under MIT. A declared, OSI-approved license is one of the transparency signals HVTrust scores.

Are Claude Agent SDK's commits signed?

16% of the last 100 commits to Claude Agent SDK are verified-signed (GPG, SSH, S/MIME, or GitHub's signing flow). Signed commits help confirm that code was authored by who the commit claims.

Not a safety endorsement. HVTracker describes what public signals show, not whether a project is safe for your use case. Run your own security review before adopting in production.

AI agent surface

MCP, providers, tool surface

Scored in HVTrust

These runtime-trust fields — detected from public repo docs and manifests — contribute a bounded adjustment to this project's HVTrust score alongside supply-chain evidence. The exact values each field can add or subtract are documented in the methodology → Compare this surface across every listed agent in the capability matrix →

MCP Server Support

medium confidence

Implemented

Claude Agent SDK appears to expose MCP server capabilities.

Detailed evidence is not shown in the public view.

External Service Dependencies

high confidence

4 detected

Public provider/service dependencies detected.

Credential signal: No explicit API-key/config marker detected.

Tool / Plugin Surface

high confidence

Declared

Declared plugin/integration surface detected.

database

Detailed evidence is not shown in the public view.

Package Provenance Drift

high confidence

Match

Published package metadata matches the tracked repo

Detailed evidence is not shown in the public view.

MCP signal live
External deps live
Tool / plugin surface live
Package provenance drift live

Maintain Claude Agent SDK?

For maintainers

HVTrust scores Claude Agent SDK from public signals only — we never contact maintainers first. If a signal is wrong, stale, or missing (provenance you publish, a Scorecard you run, signed releases), tell us and we'll review it. Corrections are public and tracked on GitHub.

Submit a correction → Add the trust badge

Reputation Timeline

Signal history

2026-10-08

Rank Moved

Rank rose 13 spots (#300 → #287)

2026-09-23

Rank Moved

Rank dropped 177 spots (#113 → #290)

2026-09-21

Rank Moved

Rank rose 177 spots (#290 → #113)

2026-09-17

Rank Moved

Rank rose 15 spots (#306 → #291)

2026-09-13

Rank Moved

Rank dropped 10 spots (#293 → #303)

2026-09-07

Rank Moved

Rank dropped 19 spots (#255 → #274)

2026-09-05

Rank Moved

Rank dropped 55 spots (#196 → #251)

2026-08-31

Rank Moved

Rank dropped 20 spots (#152 → #172)

2026-08-29

Rank Moved

Rank dropped 11 spots (#144 → #155)

2026-08-26

Rank Moved

Rank dropped 18 spots (#122 → #140)

2026-08-13

Rank Moved

Rank dropped 10 spots (#108 → #118)

2026-08-12

Rank Moved

Rank rose 14 spots (#122 → #108)

2026-07-14

Grade Changed

Trust grade D → B

2026-07-14

Rank Moved

Rank rose 216 spots (#307 → #91)

2026-07-14

HVTrust Changed

HVTrust up 28.9pts (42.6 → 71.5)

GitHub → PyPI: claude-agent-sdk

Embed Badge Badge guide for maintainers →

For maintainers

Markdown:
            [![HVTrust](https://hvtracker.net/badge/claude-agent-sdk.svg)](https://hvtracker.net/agents/claude-agent-sdk)
            
HTML:
            <a href="https://hvtracker.net/agents/claude-agent-sdk"><img src="https://hvtracker.net/badge/claude-agent-sdk.svg" alt="HVTrust"></a>

Other agents in Agent Frameworks

Data sources
GitHub REST API (repo, commits, stars, forks, license) · PyPI / pypistats (downloads, provenance) · OpenSSF Scorecard CLI
Each agent's signals refresh once daily across 6 staggered batches. Methodology v4.4 · Raw JSON

Claude Agent SDK

How Claude Agent SDK could raise its score

Chain of custody

Source Partly signed

Review and release Mixed

Published packages Linked, not attested

What you install today No known advisories

Changes to this chain

Evidence behind this score

Verify this score yourself

What this score doesn’t check

How Claude Agent SDK compares in Agent Frameworks

Where the 72.4 comes from

HVTrust dimensions vs the Agent Frameworks average

Quick Trust Read

Rank Trend

Activity & Reach

Analysis

Activity Inputs

Common questions about Claude Agent SDK

AI agent surface

Maintain Claude Agent SDK?

Reputation Timeline

Embed Badge Badge guide for maintainers →

Other agents in Agent Frameworks