{"passport":{"unfragile":{"@version":"1.0","version":"2026-05","artifact":{"id":"cline-claude-dev","slug":"cline-claude-dev","name":"Cline (Claude Dev)","type":"agent","url":"https://marketplace.visualstudio.com/items?itemName=saoudrizwan.claude-dev","page_url":"https://unfragile.ai/cline-claude-dev","categories":["code-editors"],"tags":[],"pricing":{"model":"free","free":true,"starting_price":null},"status":"active","verified":false},"capabilities":[{"id":"cline-claude-dev__cap_0","uri":"capability://code.generation.editing.autonomous.file.creation.and.editing.with.approval.gates","name":"autonomous-file-creation-and-editing-with-approval-gates","description":"Cline analyzes task descriptions and project context to autonomously generate and modify source files within the VS Code workspace. The agent uses Claude/GPT-4 reasoning to determine which files to create or edit, generates code changes, and presents them for explicit human approval before writing to disk. This human-in-the-loop pattern prevents unintended file system mutations while enabling multi-file refactoring and feature implementation in a single task loop.","intents":["I want the AI to create boilerplate files and project structure without me manually scaffolding","I need to refactor code across multiple files but want to review each change before it's applied","I want to add a new feature that requires creating several interdependent files","I need the AI to fix linter errors and missing imports across my codebase automatically"],"best_for":["solo developers building features end-to-end","teams migrating codebases who want AI-assisted refactoring with human oversight","developers prototyping new projects who want scaffolding without manual setup"],"limitations":["Each file change requires explicit user approval — cannot batch-apply changes without individual confirmation, slowing multi-file operations","No built-in conflict detection for concurrent edits — if user manually edits a file while Cline proposes changes to it, merge behavior is undefined","Context window limits mean large projects may require file selection heuristics that could miss relevant code dependencies","No transaction semantics — if approval is given for 5 file changes and one fails mid-write, partial state is left on disk"],"requires":["VS Code 1.93+ (for shell integration)","Active API key for Claude, GPT-4, or compatible provider","Write permissions to workspace directory"],"input_types":["natural language task description","project source code (read via AST analysis)","linter/compiler error output"],"output_types":["source code files (JavaScript, Python, TypeScript, etc.)","configuration files (JSON, YAML, TOML)","markup files (HTML, Markdown)"],"categories":["code-generation-editing","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_1","uri":"capability://automation.workflow.terminal.command.execution.with.output.parsing","name":"terminal-command-execution-with-output-parsing","description":"Cline can execute arbitrary shell commands in the VS Code integrated terminal, capture stdout/stderr output, and parse results to inform subsequent actions. The agent uses command output to detect build failures, test results, deployment status, and runtime errors, then reacts by proposing fixes or next steps. Each command execution requires explicit human approval before running, and the agent receives full terminal output context for decision-making.","intents":["I want the AI to run my build/test suite and fix failures it detects","I need to install dependencies and configure the environment without manual terminal commands","I want the AI to deploy my code and handle deployment errors automatically","I need the AI to run database migrations or setup scripts as part of a larger task"],"best_for":["developers building CI/CD-adjacent workflows within their editor","teams using Cline for full-stack development where build/test feedback drives code changes","developers prototyping applications that require environment setup (npm install, migrations, etc.)"],"limitations":["Commands execute in user's actual shell environment (not sandboxed) — requires explicit approval to prevent accidental destructive commands like rm -rf","No timeout enforcement documented — long-running commands could block the agent indefinitely","Output parsing relies on regex/heuristics — complex or non-standard command output may not be correctly interpreted","No environment variable isolation — commands inherit user's full shell environment, which could leak secrets if not carefully managed","Terminal state is not reset between commands — side effects from one command (e.g., cd into a directory) persist for subsequent commands"],"requires":["VS Code 1.93+ with shell integration enabled","Access to shell (bash, zsh, PowerShell, cmd.exe depending on OS)","Necessary tools installed in PATH (npm, python, git, etc. depending on task)"],"input_types":["natural language task description","project configuration files (package.json, requirements.txt, etc.)","previous command output"],"output_types":["stdout/stderr text","exit codes","file system side effects (created/modified files)"],"categories":["automation-workflow","tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_10","uri":"capability://planning.reasoning.task.loop.execution.with.iterative.refinement","name":"task-loop-execution-with-iterative-refinement","description":"Cline executes tasks as multi-step loops where each step (file edit, command execution, browser interaction) produces output that informs the next step. The agent uses feedback from previous steps to refine its approach, detect errors, and iterate toward task completion. A single task can involve dozens of steps across file operations, terminal commands, and browser interactions, with the agent maintaining context across all steps.","intents":["I want to describe a complex task and have the AI break it down and execute it step-by-step","I need the AI to iterate and fix issues as it encounters them during task execution","I want the AI to learn from command output and adjust its approach accordingly","I need multi-step workflows (e.g., create files, install dependencies, run tests, deploy) executed in sequence"],"best_for":["developers with complex, multi-step tasks that require iteration and feedback","teams building end-to-end features that span multiple files and systems","developers who want AI to handle task orchestration without manual step-by-step direction"],"limitations":["Task loop latency accumulates with each step — 20-step task could take 5-10 minutes depending on API latency and approval time","No explicit loop termination condition — agent must decide when task is complete, which may be ambiguous","Context window grows with each step — long task loops may exhaust token budget before completion","Error recovery is agent-dependent — if agent misinterprets an error, it may pursue wrong fix direction","No checkpointing — if task fails mid-loop, no way to resume from last successful step"],"requires":["Sufficient API token budget for multi-step execution","Stable API connection (network interruptions could break task loop)","VS Code 1.93+"],"input_types":["natural language task description","project context (files, configuration)"],"output_types":["completed task (files created/modified, commands executed, browser interactions performed)","task execution log (step-by-step actions taken)"],"categories":["planning-reasoning","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_11","uri":"capability://automation.workflow.vs.code.workspace.integration.with.sidebar.ui","name":"vs-code-workspace-integration-with-sidebar-ui","description":"Cline integrates into VS Code as a sidebar panel, providing a dedicated UI for task input, action approval, and execution monitoring. The sidebar displays proposed actions, token usage, and task progress, allowing developers to interact with the agent without context-switching to other tools. The extension integrates with VS Code's file explorer and terminal, enabling seamless workflow within the editor.","intents":["I want to use an AI agent without leaving VS Code","I need a dedicated UI for task input and action approval within my editor","I want to see token usage and costs without opening external dashboards","I need the AI to integrate with my existing VS Code workflow and tools"],"best_for":["developers who spend most time in VS Code and want integrated AI assistance","teams using VS Code as their primary development environment","developers who want to minimize context-switching between tools"],"limitations":["Sidebar UI real estate is limited — complex tasks with many proposed actions may be hard to review in sidebar","No remote VS Code support documented — unclear if extension works with SSH, WSL, or Codespaces","Extension size and performance impact not documented — could slow VS Code startup or responsiveness","Sidebar state is not persisted — closing VS Code loses task history and context","No keyboard shortcuts documented — may require mouse interaction for approval workflow"],"requires":["VS Code 1.93+","Extension installed from marketplace (saoudrizwan.claude-dev)"],"input_types":["natural language task description (text input in sidebar)","image uploads (screenshots, mockups)"],"output_types":["sidebar UI displaying proposed actions","approval/rejection UI","token usage and cost display"],"categories":["automation-workflow","tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_2","uri":"capability://image.visual.headless.browser.automation.with.visual.feedback","name":"headless-browser-automation-with-visual-feedback","description":"Cline can launch a headless browser instance, perform user interactions (click, type, scroll), capture screenshots and console logs, and detect visual/runtime bugs. The agent uses browser feedback to understand application behavior, identify UI issues, and propose fixes. This enables testing and debugging of web applications without leaving VS Code, with visual evidence (screenshots) informing code changes.","intents":["I want the AI to test my web app in a browser and fix visual bugs it finds","I need to verify that my frontend changes render correctly without manual browser testing","I want the AI to interact with my running dev server and report what it sees","I need to debug runtime errors that only appear in the browser, not in the terminal"],"best_for":["full-stack developers building web applications who want integrated browser testing","frontend developers prototyping UI changes and wanting automated visual verification","developers debugging browser-specific issues without context-switching to DevTools"],"limitations":["Headless-only — cannot interact with user's primary browser or test against real browser extensions/plugins","Screenshot-based visual analysis relies on Claude's image understanding — may miss subtle CSS issues or accessibility problems","No direct access to browser DevTools — cannot inspect network requests, DOM state, or performance metrics programmatically","Interaction latency — headless browser startup and navigation adds 2-5 seconds per browser session","Limited to localhost/accessible URLs — cannot test against production sites or sites behind authentication without credentials"],"requires":["Headless browser engine (Chromium/Puppeteer or similar) installed","Running dev server or accessible web application URL","VS Code 1.93+"],"input_types":["web application URL","interaction instructions (click selectors, type text, scroll)","screenshot images"],"output_types":["screenshot images","console log output","error messages and stack traces","proposed code fixes"],"categories":["image-visual","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_3","uri":"capability://data.processing.analysis.context.aware.codebase.analysis.with.ast.parsing","name":"context-aware-codebase-analysis-with-ast-parsing","description":"Cline analyzes project structure and source code using Abstract Syntax Tree (AST) parsing and regex-based file searching to understand dependencies, imports, and code relationships. The agent uses this analysis to select relevant files for context, avoiding token limit exhaustion on large projects. This enables the agent to reason about multi-file changes while staying within API token budgets.","intents":["I want the AI to understand my project structure without me manually explaining it","I need the AI to find all usages of a function or class across my codebase","I want the AI to refactor code while respecting import dependencies and avoiding breaking changes","I need the AI to work on large projects without hitting token limits"],"best_for":["developers working on large codebases (10k+ lines) where full context is infeasible","teams with complex dependency graphs who need AI-assisted refactoring","developers building multi-package monorepos who need cross-package awareness"],"limitations":["AST parsing limited to supported languages — no documentation on which languages are supported (likely JavaScript/TypeScript, Python, possibly others)","Regex-based file searching may miss complex dependency patterns (e.g., dynamic imports, metaprogramming)","Context selection heuristics are not transparent — developers cannot control which files are included/excluded from context","No incremental analysis — full AST parse required for each task, adding latency on large projects","Circular dependency detection not documented — agent may miss circular import issues"],"requires":["Project with supported language syntax","VS Code 1.93+","Reasonable project size (exact threshold unknown, but documented as handling 'large projects')"],"input_types":["source code files","project configuration (package.json, tsconfig.json, etc.)","file system structure"],"output_types":["dependency graph (implicit in reasoning)","file relevance scores (implicit)","import/export analysis"],"categories":["data-processing-analysis","memory-knowledge"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_4","uri":"capability://tool.use.integration.multi.provider.llm.orchestration.with.model.switching","name":"multi-provider-llm-orchestration-with-model-switching","description":"Cline abstracts away provider-specific API differences by supporting Claude, GPT-4, Gemini, Bedrock, Azure OpenAI, Vertex AI, Cerebras, Groq, and local models (LM Studio, Ollama) through a unified configuration interface. The agent can switch between providers and models without code changes, and when using OpenRouter, it automatically fetches the latest available model list for real-time model selection. This enables users to choose the best model for their task without vendor lock-in.","intents":["I want to use Claude for my main coding tasks but switch to GPT-4 for specific complex problems","I need to use a local model to avoid API costs and keep code private","I want to automatically use the latest available models without manually updating configuration","I need to switch providers if one API is rate-limited or experiencing downtime"],"best_for":["developers who want flexibility in model selection without vendor lock-in","teams with cost optimization requirements who want to use cheaper models for simple tasks","organizations with data privacy requirements who need local model support","developers experimenting with different models to find the best fit for their workflow"],"limitations":["Provider configuration requires manual API key setup — no built-in OAuth or credential management","Model capabilities vary significantly across providers — agent may not adapt behavior based on model capabilities (e.g., vision support, function calling)","OpenRouter model list fetching adds latency on startup — exact latency not documented","No automatic fallback if primary provider fails — requires manual provider switching","Local model performance depends on hardware — no guidance on minimum specs for LM Studio/Ollama"],"requires":["API key for at least one provider (Anthropic, OpenAI, Google, AWS, Azure, GCP, Cerebras, Groq, or OpenRouter)","For local models: LM Studio or Ollama installed and running","VS Code 1.93+"],"input_types":["provider configuration (API endpoint, API key)","model name/identifier"],"output_types":["LLM completions","function calling schemas","token usage metrics"],"categories":["tool-use-integration","planning-reasoning"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_5","uri":"capability://data.processing.analysis.token.tracking.and.cost.calculation.per.task","name":"token-tracking-and-cost-calculation-per-task","description":"Cline tracks token consumption for each API request and aggregates usage across the entire task loop, calculating estimated costs based on provider pricing. This transparency enables developers to understand API spending and optimize task complexity. Token counts are displayed in the UI and logged per request and per task completion.","intents":["I want to understand how much each task costs in API fees","I need to track token usage to optimize my workflow and reduce costs","I want to see token consumption per request to identify expensive operations","I need to report API costs to my team or client"],"best_for":["developers paying for API usage who need cost visibility","teams with API budgets who need to track spending","cost-conscious developers optimizing task complexity vs. API expense"],"limitations":["Cost calculation based on static pricing — does not account for volume discounts or dynamic pricing changes","Token counting may be approximate — depends on provider's token counting accuracy (e.g., Claude vs. GPT-4 count differently)","No budget alerts or spending limits — cannot prevent runaway costs if task loops become inefficient","Cost visibility is post-hoc — cannot estimate cost before running a task"],"requires":["Active API key with billing enabled","VS Code 1.93+"],"input_types":["API responses with token counts"],"output_types":["token count (per request and per task)","estimated cost (USD or provider currency)","cost breakdown by request"],"categories":["data-processing-analysis","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_6","uri":"capability://tool.use.integration.model.context.protocol.mcp.tool.extension","name":"model-context-protocol-mcp-tool-extension","description":"Cline supports the Model Context Protocol (MCP) standard, allowing developers to create custom tools and extend the agent's capabilities beyond file operations, terminal commands, and browser automation. Custom MCP tools are registered in configuration and invoked by the agent when relevant to the task. This enables integration with domain-specific tools (e.g., database clients, API testing frameworks, custom linters).","intents":["I want to extend Cline with custom tools specific to my tech stack","I need the AI to interact with my internal APIs or databases","I want to create domain-specific tools that the agent can use autonomously","I need to integrate Cline with my company's custom development tools"],"best_for":["teams with custom development tools who want AI integration","developers building specialized applications (e.g., ML pipelines, data processing) who need domain-specific AI capabilities","organizations with proprietary APIs or databases that need AI-assisted interaction"],"limitations":["MCP tool development requires understanding MCP specification — not documented in provided content","Tool discovery and invocation relies on agent reasoning — agent may not recognize when to use custom tools","No built-in tool versioning or compatibility checking — breaking changes to tools could break agent workflows","Tool execution happens in agent's process context — no sandboxing or resource limits","Documentation on MCP integration is minimal — developers must refer to external MCP specification"],"requires":["Understanding of Model Context Protocol (MCP) specification","Custom tool implementation (language/framework not specified)","MCP server running and accessible","VS Code 1.93+"],"input_types":["MCP tool schema definitions","tool input parameters"],"output_types":["tool execution results","structured data from custom tools"],"categories":["tool-use-integration","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_7","uri":"capability://safety.moderation.human.in.the.loop.approval.workflow.with.transparency","name":"human-in-the-loop-approval-workflow-with-transparency","description":"Every autonomous action (file write, terminal command, browser interaction) requires explicit user approval before execution. Cline presents proposed actions with full details (file paths, command text, interaction targets) and waits for user confirmation. This prevents unintended mutations while maintaining agent autonomy for reasoning and planning. Users can review, reject, or modify proposed actions before approval.","intents":["I want to use an autonomous agent but need control over what it does to my codebase","I need to review AI-proposed changes before they're applied to production code","I want to understand the agent's reasoning and see what it plans to do","I need to reject or modify proposed actions that don't match my intent"],"best_for":["teams with code review requirements who want AI assistance with human oversight","developers working on production code who cannot tolerate unreviewed changes","organizations with compliance requirements (SOC 2, ISO 27001) that mandate human approval for code changes","developers learning from AI-generated code who want to see and understand each step"],"limitations":["Approval workflow adds latency — each action requires human interaction, slowing multi-step tasks","No batch approval — cannot approve multiple similar actions at once (e.g., 'apply all linter fixes')","Approval UI/UX not documented — unclear how easy it is to review and approve complex changes","No audit trail documented — cannot verify which user approved which changes in a team setting","Rejection handling not documented — unclear if agent learns from rejected actions or simply stops"],"requires":["Active user at keyboard to approve actions","VS Code 1.93+"],"input_types":["proposed action (file path, command, interaction target)","user approval/rejection decision"],"output_types":["approval decision","modified action (if user edits before approval)"],"categories":["safety-moderation","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_8","uri":"capability://code.generation.editing.linter.and.compiler.error.detection.with.proactive.fixing","name":"linter-and-compiler-error-detection-with-proactive-fixing","description":"Cline monitors linter and compiler output (from terminal execution or file watching) and proactively detects errors like missing imports, syntax errors, and type mismatches. When errors are detected, the agent proposes fixes (adding imports, correcting syntax, adjusting types) without waiting for explicit user request. This enables a tight feedback loop where code quality issues are automatically remediated.","intents":["I want the AI to automatically fix linter warnings and errors as it generates code","I need missing imports to be added automatically without manual intervention","I want type errors to be fixed as the agent refactors code","I need the AI to maintain code quality standards (ESLint, mypy, etc.) automatically"],"best_for":["developers using strict linting/type checking who want automated compliance","teams with code quality standards who want AI to maintain them","developers building large codebases where manual error fixing is tedious"],"limitations":["Error detection depends on linter/compiler output — if linter is not configured or not run, errors may be missed","Proactive fixing may introduce false positives — agent may misinterpret error messages and apply incorrect fixes","No learning from repeated errors — agent doesn't improve its error-fixing heuristics over time","Limited to errors that produce structured output — complex or ambiguous errors may not be detected","No configuration for error severity — agent treats all errors equally rather than prioritizing critical issues"],"requires":["Linter/compiler configured in project (ESLint, TypeScript, mypy, etc.)","Linter/compiler output accessible to Cline (via terminal or file watching)","VS Code 1.93+"],"input_types":["linter/compiler error output","source code files"],"output_types":["proposed fixes (file edits)","corrected source code"],"categories":["code-generation-editing","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__cap_9","uri":"capability://image.visual.image.input.for.visual.bug.fixing.and.mockup.implementation","name":"image-input-for-visual-bug-fixing-and-mockup-implementation","description":"Cline accepts image uploads (screenshots, mockups, design files) as input to tasks, enabling developers to describe visual bugs or design requirements visually rather than textually. The agent uses Claude's vision capabilities to analyze images and generate code to match the visual specification. This enables use cases like fixing visual bugs from screenshots or implementing designs from mockups.","intents":["I want to upload a screenshot of a bug and have the AI fix it","I need to implement a design from a mockup image without manual translation to code","I want to show the AI a visual problem and have it propose a fix","I need to verify that my code matches a design screenshot"],"best_for":["frontend developers who work with designers and want to implement designs directly from mockups","developers debugging visual bugs who want to show the AI what's wrong","teams using design tools (Figma, Adobe XD) who want to bridge design-to-code gap"],"limitations":["Image analysis depends on Claude's vision capabilities — may miss subtle design details or accessibility issues","Mockup-to-code translation is approximate — generated code may not perfectly match design (spacing, colors, animations)","No design system integration — agent doesn't understand design tokens or component libraries from images alone","Image format support not documented — unclear which formats are supported (PNG, JPG, SVG, PDF, etc.)","No design versioning — if design changes, agent doesn't track previous versions or diffs"],"requires":["Image file (format not specified, likely PNG/JPG/SVG)","Claude model with vision capabilities (Claude 3 or later)","VS Code 1.93+"],"input_types":["image files (screenshots, mockups, design files)","natural language description of visual problem or design requirement"],"output_types":["source code (HTML, CSS, JavaScript, React, etc.)","proposed visual fixes"],"categories":["image-visual","code-generation-editing"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"cline-claude-dev__headline","uri":"capability://code.generation.editing.ai.coding.assistant.for.visual.studio.code","name":"ai coding assistant for visual studio code","description":"Cline is an autonomous AI coding agent that integrates with Visual Studio Code, enabling developers to create files, run commands, and automate browser tasks with human approval for each action.","intents":["best AI coding assistant","AI coding agent for VS Code","autonomous coding tools for developers","AI tools for code automation","AI agents for software development"],"best_for":["developers using VS Code","teams needing code automation"],"limitations":["requires human approval for actions"],"requires":["Visual Studio Code"],"input_types":["code files","terminal commands"],"output_types":["executed commands","automated tasks"],"categories":["code-generation-editing"],"confidence":0.5,"matches":0,"success_rate":0}],"trust":{"score":77,"verified":false,"data_access_risk":"high","permissions":["VS Code 1.93+ (for shell integration)","Active API key for Claude, GPT-4, or compatible provider","Write permissions to workspace directory","VS Code 1.93+ with shell integration enabled","Access to shell (bash, zsh, PowerShell, cmd.exe depending on OS)","Necessary tools installed in PATH (npm, python, git, etc. depending on task)","Sufficient API token budget for multi-step execution","Stable API connection (network interruptions could break task loop)","VS Code 1.93+","Extension installed from marketplace (saoudrizwan.claude-dev)"],"failure_modes":["Each file change requires explicit user approval — cannot batch-apply changes without individual confirmation, slowing multi-file operations","No built-in conflict detection for concurrent edits — if user manually edits a file while Cline proposes changes to it, merge behavior is undefined","Context window limits mean large projects may require file selection heuristics that could miss relevant code dependencies","No transaction semantics — if approval is given for 5 file changes and one fails mid-write, partial state is left on disk","Commands execute in user's actual shell environment (not sandboxed) — requires explicit approval to prevent accidental destructive commands like rm -rf","No timeout enforcement documented — long-running commands could block the agent indefinitely","Output parsing relies on regex/heuristics — complex or non-standard command output may not be correctly interpreted","No environment variable isolation — commands inherit user's full shell environment, which could leak secrets if not carefully managed","Terminal state is not reset between commands — side effects from one command (e.g., cd into a directory) persist for subsequent commands","Task loop latency accumulates with each step — 20-step task could take 5-10 minutes depending on API latency and approval time","builder identity is not verified yet","no observed match outcomes yet"],"rank_breakdown":{"adoption":0.7,"quality":0.9,"ecosystem":0.3,"match_graph":0.25,"freshness":0.75,"weights":{"adoption":0.25,"quality":0.25,"ecosystem":0.1,"match_graph":0.28,"freshness":0.12}},"observed_outcomes":{"matches":0,"success_rate":0,"avg_confidence":0,"top_intents":[],"last_matched_at":null},"maintenance":{"status":"active","updated_at":"2026-05-24T12:16:21.547Z","last_scraped_at":null,"last_commit":null},"community":{"stars":null,"forks":null,"weekly_downloads":null,"model_downloads":null,"model_likes":null}},"distribution":{"claim_url":"https://unfragile.ai/submit?claim=cline-claude-dev","compare_url":"https://unfragile.ai/compare?artifact=cline-claude-dev"}},"signature":"9F8zttOz4zRgoTsfP0aQY68xogYnL7aB7xYq1y4qTk28iXI9bnr89s90i8X3LVoJK3NHxR43/P3BOyqIBEV1DA==","signedAt":"2026-06-22T15:19:35.913Z","signedBy":"unfragile.ai","version":1},"_links":{"self":"https://unfragile.ai/api/v1/passport/cline-claude-dev","artifact":"https://unfragile.ai/cline-claude-dev","verify":"https://unfragile.ai/api/v1/verify?slug=cline-claude-dev","publicKey":"https://unfragile.ai/api/v1/trust-passport-public-key","spec":"https://unfragile.ai/trust","schema":"https://unfragile.ai/schema.json","docs":"https://unfragile.ai/docs"}}