Claude Agent SDK vs Strands Agents

Claude Agent SDK 40

No payment protocol (0). Claude model prices and Managed Agents at $0.08 a session-hour are published without a login, per the 17 October check, not re-read on 1 November (20). The package is free and needs no card (20). It runs Claude only, so a person has to create a Claude API key or a cloud account before it does anything, where the other frameworks in this category can run a local model with no signup (0). 0+20+20+0 = 40. Judgement call, and it goes against Anthropic's own listing. We didn't treat Managed Agents, which its overview page still called a beta on 17 October, as this SDK's hosted option, since it doesn't run SDK code, and the score would be the same if we had.

Strands Agents 60

Self-hosted rule. Free Apache-2.0 packages with nothing to buy from the project (20 + 20 + 20). No payment protocol (0). Model calls are paid to the provider. 0+20+20+20 = 60.

Claude Agent SDK 70

Official packages on PyPI (Requires-Python >=3.10 in pyproject.toml, 0.2.176 tagged on 31 October) and npm (TypeScript 0.3.311 on 26 October raised the minimum Node.js from 18.0.0 to 18.14.0) (20). The Python repository still runs its Test and Lint workflows, and 0.2.169 and 0.2.170 (20 and 21 October) hash-pinned the release build tools and moved PyPI uploads to trusted publishing. Runs on main pass per the 1 October check, not re-read on 1 November (25). 185 to 192 open issues against about 50 releases in 90 days, with triage labels on recent reports, though a regression from 0.2.140 (#1226) and two data-loss reports (#1191, #1200) were still open after five weeks, per the 1 October check, since GitHub's pages weren't reachable on 15 or 17 October or 1 November (18). The changelog has Breaking Changes headings but no dates, 0.2.129 broke skill names in a patch-level bump, TypeScript 0.3.286, 0.3.300 and 0.3.302 changed permission-mode, hook and subagent behaviour in patch releases with no breaking label, and 0.3.311 on 26 October raised the Node.js floor the same way (7). Python 0.2.176 and TypeScript 0.3.316 are pre-1.0, and pyproject.toml still classes the package as 3 - Alpha (0). 20+25+18+7+0 = 70. Judgement call. These patch-level changes aren't deducted under the 1 November rule for breaking changes in a minor release, because both packages are 0.x, the provisional exemption bird v0.77.0 got.

Strands Agents 87

Local-package reading. Official PyPI and npm packages with Python 3.10 and Node 22 stated (20). A public CI workflow runs lint, type checks, unit and integration tests for each package on pull requests to main, plus security-audit workflows, and the latest Python, TypeScript, harness and MCP runs on main passed on 6 November (25). 664 open issues and 1,151 open issues and pull requests on 6 November, with 545 issues opened in the 90 days to 3 November. Of the 50 newest, 12 had a comment and 15 were closed, and 9 of the 40 newest open items carry a bug label. An issue-responder workflow runs on new issues (12). A written semantic versioning policy, experimental APIs kept apart, and deprecations removed only at a major (15). 1.0 since 15 July 2025 (15). 20+25+12+15+15 = 87.

Claude Agent SDK 83

The Python wrapper is MIT, but its wheel bundles the proprietary Claude Code binary, and the TypeScript package's LICENSE.md still says All rights reserved under Anthropic's Commercial Terms, both re-read in the clones on 1 November (18). The data-usage page, re-read on 1 November, states 30-day retention for commercial and API users, zero data retention for qualified accounts, no training on commercial data and 30 days of local transcripts in ~/.claude/projects, but still names the logging and error-tracking services only as third parties (25). Deprecations appear in the changelogs without dates, and there's no written deprecation policy for the SDK (8). Telemetry is documented per provider, with an opt-out for each service (20). 18+25+8+20 = 71. Judgement call on the licence line, written on Claude.

Strands Agents 72

Apache-2.0 (30). A library that stores session state where the developer configures it. The default FileSessionManager kept sessions world-readable in /tmp until 1.48.0, which the October advisories record and fix (20). A written deprecation policy with warnings and removal at majors, without a dated support window (16). Telemetry is opt-in and documented (20). 30+20+16+20 = 86.

Claude Agent SDK 95

The documented agent with one MCP server is about 15 lines with the built-in client, and tool search, the default, withholds MCP tool definitions, both re-read on the MCP page on 1 November (25). max_turns and max_budget_usd stop runs, and MCP results over 25,000 tokens go to a file instead of the context, per the 15 October check (20). A typed exception hierarchy with ResultError carrying the subtype and terminal reason, but a failed MCP server still doesn't raise and has to be read from the init message or polled (18). Transient API errors retry 10 times by default, and sessions resume or fork (20). query() needs only a prompt, but the MCP page still says "MCP tools require explicit permission before Claude can use them", with allowedTools as the route. Since Claude Code 2.1.302, which Python 0.2.176 and TypeScript 0.3.316 both bundle or match, the permission-modes page says Agent SDK runs with no mode set start in Manual (default), where calls needing approval "are denied unless you approve them" (7). Python and TypeScript (5). 25+20+18+20+7+5 = 95. Judgement call. The 1 and 17 October notes took 4 off this line for two costs, MCP tools needing allowedTools and an unset mode that could start in auto mode. The second has gone. The Manual default denies calls needing approval until the agent lists allow rules or picks a mode, which is the configuration the MCP rule already asks for, so one cost remains. Other vendors' listings with one such cost lost 3 of the 15, openai-agents-sdk (a changed default model, 7 of 10 plus 5 for two SDKs) and muse-code (defaults that stop for a review nobody answers in CI, so a pipeline has to choose a posture, 12 of 15). We took that reading. Keeping 6 would count one cost twice, and the lighter reading gives 8.

Strands Agents 87

Framework reading. An MCPClient is passed straight to the agent in a few lines and its lifecycle is managed, with tool filtering per client (22). Conversation managers, turn limits, token budgets and cancellation bound each run (20). Stop reasons and typed events, and hooks that can retry a model or tool call (16). Retry strategies, session managers that persist and resume a conversation, and interrupts that resume where they paused (17). Two official languages, but the default provider is Bedrock, which needs AWS credentials and model access (12). 22+20+16+17+12 = 87.

Claude Agent SDK 89

Typed public interfaces with a full Python and TypeScript reference, options as dataclasses and TypedDicts (25). llms.txt at code.claude.com/docs/llms.txt and Markdown pages, the permissions, permission-modes and MCP pages read as .md on 1 November (10). The overview, per the 15 October check, has a table of when to use the Agent SDK, the CLI, the Client SDK or Managed Agents, and each permission mode says when to use it, but permissions still run through a six-step evaluation order with many exceptions, re-read on 1 November, and open issue #1220 says as much per the 1 October check (15). Permission modes and MCP server configs are typed variants, Python 0.2.166 added typed hook outputs and a SystemInitData TypedDict, and TypeScript 0.3.312 to 0.3.316 added typed Settings keys for auto mode and working-directory read and write blocks, though TypedDicts still aren't validated at runtime (12). Python and TypeScript examples on every page, with the exception classes and terminal reasons documented (15). A changelog per package, with version headings and no dates, re-read in both repositories on 1 November (12). 25+10+15+12+15+12 = 89.

Strands Agents 95

Framework reading. Typed Python and TypeScript APIs with a reference for each (25). llms.txt with a Markdown twin of every page (10). The llms.txt and docs say when to use Strands over a hand-written loop and when to take the harness rather than the SDK (18). Tools are typed from decorated functions or Zod schemas (15). Python and TypeScript examples on every page, stop reasons and hook events documented, with no catalogue of exceptions (12). Versioning page and tagged releases per package (15). 25+10+18+15+12+15 = 95.

Claude Agent SDK 83

Usage metrics go to Anthropic and third-party logging by default on the Claude API, without code, prompts or file paths, and DISABLE_TELEMETRY=1 turns them off. They're off by default on Google Cloud, Bedrock, Foundry and Claude Platform on AWS, unless a host platform sets CLAUDE_CODE_PROVIDER_MANAGED_BY_HOST. WebFetch sends hostnames to api.anthropic.com for a blocklist check on every provider, with its own opt-out. All re-read on the data-usage page on 1 November (20). canUseTool approval callback and PreToolUse hooks, plan and dontAsk modes with deny rules, and sandbox settings for shell commands, re-read on the permissions page on 1 November. TypeScript 0.3.301 added per-tool always_ask and always_deny policies for in-process MCP servers, and 0.3.314 and 0.3.316 typed blockWritesOutsideWorkingDirectories and blockReadsExemptPaths (20). Hooks can block a call and auto mode runs a classifier on actions, but there's no input or output guardrail primitive (10). OpenTelemetry export to your own collector stays off until CLAUDE_CODE_ENABLE_TELEMETRY=1 is set, prompt and tool content need separate flags, and tool_decision and permission_mode_changed events can form a per-user audit trail, per the observability page read on 15 October (15). HackerOne bug bounty, advisories published on the claude-code repository, whose SECURITY.md is unchanged in the clone, and anthropic.com's security.txt valid to 31 December 2026, re-read on 1 November, but neither the Python nor the TypeScript SDK repository has a SECURITY.md, checked again in both clones on 1 November (18). 20+20+10+15+18 = 83. Framework reading, so SOC 2 isn't scored, as for claude-code. The bundled CLI's advisories are counted under negative events, not here.

Strands Agents 89

Framework reading (telemetry defaults, approvals, guardrails, sandboxing). No usage analytics found in either SDK, and OpenTelemetry is opt-in (25). Interrupts and the harness's interventions gate tool calls behind human approval or Cedar policies, but the default sandbox is the local host (16). Guardrails, PII redaction and trusted message history pages, with Bedrock Guardrails integration (13). OpenTelemetry traces for every model and tool call (15). The AWS Vulnerability Disclosure Program on HackerOne and three advisories in October 2026 published with fixed versions (20). 25+16+13+15+20 = 89.