Python imports + 34 CLIs¶

The most direct way to use vstack: import the pattern and call it.

from vstack.lewin import LewinAttributionDetector, AgentFailureTrace, FailureStep
from vstack.aar import AnthropicClient

detector = LewinAttributionDetector(AnthropicClient(), mode="standard")
detection = detector.run(
    AgentFailureTrace(
        agent_id="qa-bot",
        model_name="claude-opus-4-7",
        task="Answer 'When was Pluto reclassified?'",
        steps=[
            FailureStep(type="input",       content="When was Pluto reclassified?"),
            FailureStep(type="tool_call",   content="rag.search(query='pluto')"),
            FailureStep(type="observation", content="returned a 2003 Wikipedia revision"),
            FailureStep(type="output",      content="Pluto was reclassified in 2003."),
        ],
        outcome="Confidently wrong year (correct: 2006).",
        success=False,
        initial_attribution="model is bad at facts",
    )
)
print(detection.to_markdown())

Every pattern follows the same interface — see The 5-layer pattern shape.

34 per-pattern CLIs¶

Every pattern also ships a CLI registered in [project.scripts]. The full list:

CLI	Pattern
`vstack`	`vstack.aar` (foundational AAR)
`vstack-lewin`	`#01 Lewin attribution`
`vstack-goleman`	`#02 Goleman EI`
`vstack-johari`	`#03 Johari window`
`vstack-danva`	`#04 DANVA emotion reader`
`vstack-reappraisal`	`#05 Cognitive reappraisal`
`vstack-yerkes`	`#06 Yerkes-Dodson`
`vstack-hexaco`	`#07 HEXACO personality`
`vstack-grant`	`#08 Grant strengths`
`vstack-motivation`	`#09 Motivation traps`
`vstack-sdt`	`#10 SDT`
`vstack-mcgregor`	`#11 McGregor`
`vstack-vroom`	`#12 Vroom`
`vstack-grpi`	`#13 GRPI`
`vstack-process`	`#14 Process gain/loss`
`vstack-loafing`	`#15 Social loafing`
`vstack-superflocks`	`#16 Superflocks`
`vstack-lencioni`	`#17 Lencioni`
`vstack-trust-triangle`	`#18 Trust Triangle`
`vstack-mcallister`	`#19 McAllister trust`
`vstack-psych-safety`	`#20 Edmondson`
`vstack-glaser`	`#21 Glaser`
`vstack-feedback-triggers`	`#22 Stone-Heen`
`vstack-plus-delta`	`#23 Plus/Delta`
`vstack-smart-goal`	`#24 SMART goal`
`vstack-group-decision`	`#25 Group decision`
`vstack-debate-pathology`	`#26 Debate pathology`
`vstack-bias-stack`	`#27 Bias stack`
`vstack-devils-advocate`	`#28 Devil's advocate`
`vstack-thomas-kilmann`	`#29 Thomas-Kilmann`
`vstack-schein-culture`	`#31 Schein`
`vstack-robbins-culture`	`#32 Robbins-Judge`
`vstack-org-structure`	`#33 Org-structure`
`vstack-span-of-control`	`#34 Span-of-control`

Each CLI has the same 7 subcommands: analyze, batch, replay, validate, schema, playbooks, compose.

vstack-lewin analyze --trace failure.json --mode forensic
vstack-lewin playbooks   # browse the failure-mode playbooks
vstack-lewin compose     # show the cross-pattern composition manifest
vstack-lewin schema --target trace   # print the trace JSON schema