1 view

strands-agents/harness-sdk

Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud.

Python★ 8,321+455 that day#8 on trending

Analyzed September 25, 2026 at 7ae759d (main) with jscpd 5.4.0 in 1.4s. Code duplication stayed at 1.59% across 2 appearances since Sep 24.

D53.9/100

Project health490,157 lines of code (L)

Duplication57

7% in python, typescript, astro, tsx, javascript, bash, perl

Dead code64

4.9%

Complexity44

60.2%

The score is jscpd's own --health model: markup, data and text left out, tests kept. The numbers below are the trending scan's.

1.59%duplicated code · 3,224 of 202,823 lines

260clones in code

1,000code files of 2,448 scanned

7.14%duplicated over all files · 3,117 clones

Most complex files

strands-cli/src/tui/view/setup-wizard/index.tsx955 cx
strands-cli/src/tui/chat/controller.ts529 cx
strands-ts/src/models/__tests__/bedrock.test.ts438 cx
strands-cli/src/tui/view/app.tsx432 cx
strands-py/src/strands/tools/mcp/mcp_client.py413 cx
strands-py/tests/strands/models/test_bedrock.py397 cx
strands-py/src/strands/models/bedrock.py346 cx
strands-ts/src/models/bedrock.ts339 cx
strands-ts/src/agent/agent.ts323 cx
strands-cli/test/tui-chat.test.ts309 cx

Dead code 4.9% unused · 317 findings

strands-cli/src/tui/view/app.tsxunused-file
strands-cli/src/tui/view/frog-intro-renderer.tsunused-file
site/src/content/docs/user-guide/sdk/agents/hooks.tsunused-file
strands-cli/src/tui/view/chat-view.tsxunused-file
strands-cli/src/tui/view/panels.tsxunused-file
strands-cli/src/tui/view/custom-theme-editor.tsxunused-file
site/src/content/docs/user-guide/sdk/tools/tools.tsunused-file
strands-cli/src/tui/view/frog-drawing.tsunused-file
site/src/components/community/CoursesSection.astrounused-file
strands-cli/src/tui/view/settings-panel.tsxunused-file

The code number leaves out tests, fixtures, docs, examples, data files, styles, translations and vendored code.

Largest clones in the code

37 lines in 2 placespython462 tokens

strands-py/src/strands/agent/agent.py:921–957strands-py/src/strands/agent/agent.py:1264–1300

921This method implements the conversational interface with multiple input patterns:
922- String input: Simple text input
923- ContentBlock list: Multi-modal content blocks
924- Message list: Complete messages with roles
925- No input: Use existing conversation history
926
927Args:
928    prompt: User input in various formats:
929        - str: Simple text input
930        - list[ContentBlock]: Multi-modal content blocks

First 10 of 37 lines.

36 lines in 2 placespython439 tokens

strands-py/src/strands/agent/agent.py:831–866strands-py/src/strands/agent/agent.py:925–960

831- No input: `agent()` - uses existing conversation history
832
833Args:
834    prompt: User input in various formats:
835        - str: Simple text input
836        - list[ContentBlock]: Multi-modal content blocks
837        - list[Message]: Complete messages with roles
838        - None: Use existing conversation history
839    invocation_state: Additional parameters to pass through the event loop.
840    structured_output_model: Pydantic model type(s) for structured output (overrides agent default).

First 10 of 36 lines.

31 lines in 2 placespython351 tokens

strands-py/src/strands/tools/mcp/mcp_client.py:1022–1052strands-py/src/strands/tools/mcp/mcp_client.py:1087–1117

1022This method automatically uses task-augmented execution when the client
1023opted in via ``tasks_config`` and the server advertises task support (on
1024the mcp 1.x line, the tool's ``taskSupport`` setting is also honored).
1025
1026Args:
1027    tool_use_id: Unique identifier for this tool use
1028    name: Name of the tool to call
1029    arguments: Optional arguments to pass to the tool
1030    read_timeout_seconds: Optional timeout for the tool call. On the mcp 2.x line, the timeout
1031        bounds each request round of a multi round-trip tool call rather than the call as a

First 10 of 31 lines.

7 more

32 lines in 2 placespython250 tokens
strands-py/src/strands/agent/agent.py:1149–1180 strands-py/src/strands/experimental/bidi/agent/agent.py:242–273
42 lines in 2 placespython227 tokens
strands-py/src/strands/models/llamacpp.py:500–541 strands-py/src/strands/models/openai.py:557–601
13 lines in 3 placespython106 tokens
strands-py/src/strands/models/anthropic.py:1021–1033 strands-py/src/strands/models/llamaapi.py:450–462 strands-py/src/strands/models/model.py:234–246
9 lines in 5 placespython51 tokens
strands-py/src/strands/models/llamaapi.py:456–464 strands-py/src/strands/models/model.py:240–248 strands-py/src/strands/models/ollama.py:386–394 strands-py/src/strands/models/openai.py:846–854 strands-py/src/strands/models/openai_responses.py:514–522
13 lines in 3 placespython93 tokens
strands-py/src/strands/models/llamaapi.py:366–378 strands-py/src/strands/models/llamacpp.py:623–635 strands-py/src/strands/models/ollama.py:321–333
36 lines in 2 placespython180 tokens
strands-py/src/strands/models/openai.py:550–585 strands-py/src/strands/models/openai_responses.py:849–884
6 lines in 4 placespython60 tokens
strands-py/src/strands/models/bedrock.py:1343–1348 strands-py/src/strands/models/litellm.py:339–344 strands-py/src/strands/models/model.py:276–281 strands-py/src/strands/models/openai.py:497–502

Largest clones over all files, tests and data included

88 lines in 4 placesjsondocs and examples321 tokens
site/docs/examples/cdk/deploy_to_apprunner/cdk.json:1–88 site/docs/examples/cdk/deploy_to_ec2/cdk.json:1–88 site/docs/examples/cdk/deploy_to_fargate/cdk.json:1–88 site/docs/examples/cdk/deploy_to_lambda/cdk.json:1–88
```
1{
2  "app": "npx tsx bin/cdk-app.ts",
3  "watch": {
4    "include": ["**"],
5    "exclude": [
6      "README.md",
7      "cdk*.json",
8      "**/*.d.ts",
9      "**/*.js",
10      "tsconfig.json",
```
First 10 of 88 lines.

170 lines in 2 placestypescripttests739 tokens

strands-ts/src/models/__tests__/bedrock.test.ts:3295–3464strands-ts/src/models/__tests__/bedrock.test.ts:3731–3900

3295    })
3296  })
3297})
3298
3299describe('media blocks in tool results', () => {
3300  const mockConverseStreamCommand = vi.mocked(ConverseStreamCommand)
3301
3302  it('formats image block in tool result', async () => {
3303    const provider = new BedrockModel()
3304    const imageBytes = new Uint8Array([1, 2, 3])

First 10 of 170 lines.

12 lines in 5 placestypescripttests178 tokens

strands-ts/src/context-manager/strategies/__tests__/offload-strategy.test.ts:346–357strands-ts/src/context-manager/strategies/__tests__/offload-strategy.test.ts:367–378strands-ts/src/context-manager/strategies/__tests__/offload-strategy.test.ts:388–399strands-ts/src/context-manager/strategies/__tests__/offload-strategy.test.ts:558–569strands-ts/src/context-manager/strategies/__tests__/offload-strategy.test.ts:582–593

346it('drop("assistantText") repairs alternation after removing assistant messages', async () => {
347  const messages = [
348    new Message({ role: 'user', content: [new TextBlock('q1')] }),
349    new Message({ role: 'assistant', content: [new TextBlock('a1')] }),
350    new Message({ role: 'user', content: [new TextBlock('q2')] }),
351    new Message({ role: 'assistant', content: [new TextBlock('a2')] }),
352    new Message({ role: 'user', content: [new TextBlock('q3')] }),
353    new Message({ role: 'assistant', content: [new TextBlock('a3')] }),
354    new Message({ role: 'user', content: [new TextBlock('q4')] }),
355    new Message({ role: 'assistant', content: [new TextBlock('a4')] }),

First 10 of 12 lines.

6 lines in 9 placestypescripttests87 tokens
strands-ts/src/models/openai/__tests__/chat.test.ts:1437–1442 strands-ts/src/models/openai/__tests__/chat.test.ts:1505–1510 strands-ts/src/models/openai/__tests__/chat.test.ts:1542–1547 strands-ts/src/models/openai/__tests__/chat.test.ts:1580–1585 strands-ts/src/models/openai/__tests__/chat.test.ts:1615–1624and 4 more
27 lines in 6 placespythondocs and examples126 tokens
site/docs/examples/evals-sdk/coherence_evaluator.py:7–33 site/docs/examples/evals-sdk/correctness_evaluator.py:7–32 site/docs/examples/evals-sdk/faithfulness_evaluator.py:7–32 site/docs/examples/evals-sdk/goal_success_rate_evaluator.py:7–32 site/docs/examples/evals-sdk/harmfulness_evaluator.py:7–32and 1 more

By format, code only

Format	Files	Lines	Clones	Duplicated lines	Duplication
python	337	78,508	182	2,337	2.98%
typescript	432	88,845	52	649	0.73%
css	45	5,155	15	132	2.56%
html	48	2,117	7	69	3.26%
tsx	33	10,342	3	26	0.25%
javascript	40	4,130	1	11	0.27%
astro	50	12,182	0	0	0%
bash	14	1,370	0	0	0%
perl	1	174	0	0	0%

By format, all files

Format	Files	Lines	Clones	Duplicated lines	Duplication
typescript	962	232,681	1,610	17,897	7.69%
python	757	234,338	1,224	16,185	6.91%
yaml	210	11,062	172	2,974	26.88%
markdown	118	22,415	53	858	3.83%
json	53	4,108	21	656	15.97%
css	49	6,433	17	156	2.42%
html	48	2,117	7	69	3.26%
javascript	55	5,551	5	100	1.8%
tsx	33	10,342	3	26	0.25%
markup	41	1,528	3	40	2.62%

Trending appearances

Day	Rank	Stars	Files	Lines	Clones	All files	Code	Health	Commit
Sep 25, 2026	#8	8,321 (+455)	2,448	546,401	3,117	7.14%	1.59%	D 53.9	`7ae759d`
Sep 24, 2026	#7	7,943 (+115)	2,448	546,494	3,117	7.13%	1.59%	D 54.5	`43b15b7`

Want this for your own project? Install jscpd and run jscpd ., or measure it the way the baselines were measured.