HarnessSource

Browse

Search and filter by type across the catalog

854 results in Identity, Infrastructure, Evals, Sub-Agents · page 28 of 36

Harness

Sub-AgentsPreview

read-only-auditor

Use this agent when you need a security audit that is guaranteed to make no changes to the codebase. This agent has hooks in its frontmatter that block all Write, Edit, and Bash tool calls for the duration of the audit — enforcing read-only mode at the hook level, not just by convention. Invoke for compliance reviews, pre-merge audits, or any situation where auditability and non-interference are required.\n\n<example>\nContext: A compliance officer needs a security review of payment processing code without any risk of accidental modification.\nuser: \"Audit src/payments/ for PCI-DSS compliance issues. Don't touch anything.\"\nassistant: \"I'll run the read-only-auditor on src/payments/. My frontmatter hooks block Write, Edit, and Bash for the duration of this session, so no files can be modified regardless of what I find. I'll check for: unencrypted PAN storage, logging of card data, insecure TLS configurations, and missing input validation on payment fields.\"\n<commentary>\nUse read-only-auditor when the non-modification guarantee needs to be enforced at the system level, not just trusted by convention.\n</commentary>\n</example>

UnrankedNo signals yet

Harnessclaude

securitysubagents

Sub-AgentsPreview

readme

Language Specialists are your expert guides for specific programming languages and their ecosystems. These subagents bring deep knowledge of language idioms, best practices, performance optimization t

UnrankedNo signals yet

Harnessclaude

programming-languagessubagents

Sub-AgentsPreview

readme-generator

Use this agent when you need a maintainer-ready README built from exact repository reality, with deep codebase scanning, zero hallucination, and optional git commit/push only when explicitly requested.

UnrankedNo signals yet

Harnessclaude

developer-experiencesubagent

Sub-AgentsPreview

readme-ja

Use the README_JA sub-agent for the matching task.

UnrankedNo signals yet

Harnessclaude

dlezosubagents

Sub-AgentsPreview

receipt-verifier

Expert in Ed25519 signed receipts, JCS canonicalization, hash chains, and offline verification. Use when you need to verify receipt authenticity, audit a receipt chain, detect tampering, or explain why verification failed.

UnrankedNo signals yet

Harnessclaude

agentssubagents

Sub-AgentsPreview

refactor-cleaner

Dead code cleanup and consolidation specialist. Use PROACTIVELY for removing unused code, duplicates, and refactoring. Runs analysis tools (knip, depcheck, ts-prune) to identify dead code and safely removes it.

UnrankedNo signals yet

Harnessclaudecodexcursorgeminiopencode

subagent

Sub-AgentsPreview

refactoring-specialist

Use when you need to transform poorly structured, complex, or duplicated code into clean, maintainable systems while preserving all existing behavior.

UnrankedNo signals yet

Harnessclaude

developer-experiencesubagent

Sub-AgentsPreview

reference-builder

Creates exhaustive technical references and API documentation. Generates comprehensive parameter listings, configuration guides, and searchable reference materials. Use PROACTIVELY for API docs, configuration references, or complete technical specifications.

UnrankedNo signals yet

Harnessclaude

agentssubagents

Sub-AgentsPreview

refine-issue

Refine the requirement or issue with Acceptance Criteria, Technical Considerations, Edge Cases, and NFRs

UnrankedNo signals yet

Harnessclaude

expert-advisorssubagents

Sub-AgentsPreview

reinforcement-learning-engineer

Use when designing RL environments, training agents with reward optimization, implementing policy gradient methods, or deploying decision-making systems for robotics, gaming, and autonomous operations.

UnrankedNo signals yet

Harnessclaude

data-aisubagent

InfrastructurePreview

render

Unified cloud for deploying web services, background workers, cron jobs, and databases for agent backends.

UnrankedNo signals yet

Harnessclaudecursorcodexopencodegemini

infrastructuredeploy

No one-command install · SourceDetails

InfrastructurePreview

replicate

API platform for running ML models in the cloud. Useful for deploying custom model components in agent pipelines.

UnrankedNo signals yet

Harnessclaudecursorcodexopencodegemini

infrastructureml-inference

No one-command install · SourceDetails

Sub-AgentsPreview

report-generator

Use this agent when you need to transform synthesized research findings into a comprehensive, well-structured final report. This agent excels at creating readable narratives from complex research data, organizing content logically, and ensuring proper citation formatting. It should be used after research has been completed and findings have been synthesized, as the final step in the research process. Examples: <example>Context: The user has completed research on climate change impacts and needs a final report. user: 'I've gathered all this research on climate change effects on coastal cities. Can you create a comprehensive report?' assistant: 'I'll use the report-generator agent to create a well-structured report from your research findings.' <commentary>Since the user has completed research and needs it transformed into a final report, use the report-generator agent to create a comprehensive, properly formatted document.</commentary></example> <example>Context: Multiple research threads have been synthesized and need to be presented cohesively. user: 'We have findings from 5 different researchers on AI safety. Need a unified report.' assistant: 'Let me use the report-generator agent to create a cohesive report that integrates all the research findings.' <commentary>The user needs multiple research streams combined into a single comprehensive report, which is exactly what the report-generator agent is designed for.</commentary></example>

UnrankedNo signals yet

Harnessclaude

deep-research-teamsubagents

Sub-AgentsPreview

requirements-analyst

PROACTIVELY USE this agent when you need to analyze and formalize business requirements, user needs, or high-level project ideas into detailed functional and non-functional specifications. This agent MUST BE USED for requirements analysis and specification creation tasks. This includes conducting stakeholder analysis, gathering business context, translating ambiguous requirements into clear specifications, and creating requirements documentation. Examples: <example>Context: User wants to build a new e-commerce platform but has only high-level ideas. user: 'I want to create an online store that can handle thousands of customers and integrate with payment systems' assistant: 'I'll use the requirements-analyst agent to analyze your business needs and create detailed functional requirements.' <commentary>Since the user has high-level business needs that need to be analyzed and formalized into detailed requirements, use the requirements-analyst agent.</commentary></example> <example>Context: User has a vague idea for a mobile app and needs help defining what it should actually do. user: 'I want to make an app that helps people manage their daily tasks better' assistant: 'Let me use the requirements-analyst agent to help you define the specific features and requirements for your task management application.' <commentary>The user has an abstract concept that needs to be broken down into concrete, actionable requirements.</commentary></example>

UnrankedNo signals yet

Harnessclaude

dlezosubagents

Sub-AgentsPreview

requirements-validator

PROACTIVELY USE this agent when you need to validate requirements for completeness, consistency, feasibility, and testability before moving to the design or development phase. This agent MUST BE USED for requirements validation and quality assurance tasks. Examples: <example>Context: User has drafted requirements but wants to ensure they're complete and consistent before development. user: 'I've written down all the requirements for my project. Can you review them to make sure nothing is missing?' assistant: 'I'll use the requirements-validator agent to perform a comprehensive review of your requirements for completeness and consistency.' <commentary>Since the user needs requirements validation and quality assurance, use the requirements-validator agent.</commentary></example> <example>Context: User is preparing to start development and wants to ensure their requirements meet industry standards. user: 'Before we start coding, I want to make sure our requirements are solid and follow best practices' assistant: 'I'll use the requirements-validator agent to validate your requirements against SMART criteria and requirements engineering best practices' <commentary>The user needs professional requirements validation, so use the requirements-validator agent.</commentary></example>

UnrankedNo signals yet

Harnessclaude

dlezosubagents

Sub-AgentsPreview

research-analyst

Use this agent when you need comprehensive research across multiple sources with synthesis of findings into actionable insights, trend identification, and detailed reporting.

UnrankedNo signals yet

Harnessclaude

research-analysissubagent

Sub-AgentsPreview

research-brief-generator

Use this agent when you need to transform a user's research query into a structured, actionable research brief that will guide subsequent research activities. This agent takes clarified queries and converts them into comprehensive research plans with specific questions, keywords, source preferences, and success criteria. <example>Context: The user has asked a research question that needs to be structured into a formal research brief.\nuser: "I want to understand the impact of AI on healthcare diagnostics"\nassistant: "I'll use the research-brief-generator agent to transform this query into a structured research brief that will guide our research."\n<commentary>Since we need to create a structured research plan from the user's query, use the research-brief-generator agent to break down the question into specific sub-questions, identify keywords, and define research parameters.</commentary></example><example>Context: After query clarification, we need to create a research framework.\nuser: "How are quantum computers being used in drug discovery?"\nassistant: "Let me use the research-brief-generator agent to create a comprehensive research brief for investigating quantum computing applications in drug discovery."\n<commentary>The query needs to be transformed into a structured brief with specific research questions and parameters, so use the research-brief-generator agent.</commentary></example>

UnrankedNo signals yet

Harnessclaude

deep-research-teamsubagents

Sub-AgentsPreview

research-coordinator

Use this agent when you need to strategically plan and coordinate complex research tasks across multiple specialist researchers. This agent analyzes research requirements, allocates tasks to appropriate specialists, and defines iteration strategies for comprehensive coverage. <example>Context: The user has asked for a comprehensive analysis of quantum computing applications in healthcare. user: "I need a thorough research report on how quantum computing is being applied in healthcare, including current implementations, future potential, and technical challenges" assistant: "I'll use the research-coordinator agent to plan this complex research task across our specialist researchers" <commentary>Since this requires coordinating multiple aspects (technical, medical, current applications), use the research-coordinator to strategically allocate tasks to different specialist researchers.</commentary></example> <example>Context: The user wants to understand the economic impact of AI on job markets. user: "Research the economic impact of AI on job markets, including statistical data, expert opinions, and case studies" assistant: "Let me engage the research-coordinator agent to organize this multi-faceted research project" <commentary>This requires coordination between data analysis, academic research, and current news, making the research-coordinator ideal for planning the research strategy.</commentary></example>

UnrankedNo signals yet

Harnessclaude

deep-research-teamsubagents

Sub-AgentsPreview

research-orchestrator

Use this agent when you need to coordinate a comprehensive research project that requires multiple specialized agents working in sequence. This agent manages the entire research workflow from initial query clarification through final report generation. <example>Context: User wants to conduct thorough research on a complex topic. user: "I need to research the impact of quantum computing on cryptography" assistant: "I'll use the research-orchestrator agent to coordinate a comprehensive research project on this topic" <commentary>Since this is a complex research request requiring multiple phases and specialized agents, the research-orchestrator will manage the entire workflow.</commentary></example> <example>Context: User has a vague research request that needs clarification and systematic investigation. user: "Tell me about AI safety" assistant: "Let me use the research-orchestrator to coordinate a structured research process on AI safety" <commentary>The broad nature of this query requires orchestration of multiple research phases, making the research-orchestrator the appropriate choice.</commentary></example>

UnrankedNo signals yet

Harnessclaude

deep-research-teamsubagents

Sub-AgentsPreview

research-synthesizer

Use this agent when you need to consolidate and synthesize findings from multiple research sources or specialist researchers into a unified, comprehensive analysis. This agent excels at merging diverse perspectives, identifying patterns across sources, highlighting contradictions, and creating structured insights that preserve the complexity and nuance of the original research while making it more accessible and actionable. <example>Context: The research-orchestrator has completed Phase 4 parallel research on 'LLM fine-tuning costs' using academic-researcher, web-researcher, and data-analyst. user: "Synthesize the research outputs." assistant: "I'll invoke the research-synthesizer agent to merge all specialist findings into a unified analysis." <commentary>The orchestrator has confirmed all three researcher outputs exist as files, making this the correct trigger point for synthesis. The agent will locate each output file, extract claims, and produce both synthesis-summary.md and synthesis.json.</commentary></example> <example>Context: The research-orchestrator has completed parallel research on 'WASM adoption in server-side runtimes' using academic-researcher, web-researcher, and technical-researcher. All three output files are confirmed present. user: "All researchers are done. Synthesize everything into a report." assistant: "Let me use the research-synthesizer agent to consolidate the three specialist outputs into a structured synthesis." <commentary>Three distinct researcher outputs referencing the same topic are present; the synthesis agent is the correct next step to unify them and surface contradictions and shared themes.</commentary></example>

UnrankedNo signals yet

Harnessclaude

deep-research-teamsubagents

Sub-AgentsPreview

research-technical-spike

Systematically research and validate technical spike documents through exhaustive investigation and controlled experimentation.

UnrankedNo signals yet

Harnessclaude

expert-advisorssubagents

Sub-AgentsPreview

resource-monitor

PROACTIVELY USE this agent when you need to monitor Claude Code resource usage, analyze API consumption patterns, track token usage and costs, identify resource-intensive operations, or optimize development efficiency. This agent MUST BE USED for resource monitoring and optimization tasks. Examples: <example>Context: User is concerned about Claude Code usage efficiency and costs. user: 'We've been using Claude Code extensively. Can you help us understand our usage patterns and optimize costs?' assistant: 'I'll use the resource-monitor agent to analyze our usage patterns and suggest optimization strategies.' <commentary>Since resource usage optimization is needed, use the resource-monitor agent to provide analytics and efficiency recommendations.</commentary></example> <example>Context: Development team wants to understand which operations consume the most tokens. user: 'Our Claude Code bills have been higher than expected. Can you identify what's driving the costs?' assistant: 'Let me use the resource-monitor agent to analyze your usage patterns and identify cost drivers.' <commentary>The user needs resource usage analysis, so use the resource-monitor agent to provide detailed consumption analytics.</commentary></example>

UnrankedNo signals yet

Harnessclaude

dlezosubagents

Sub-AgentsPreview

reverse-engineer

Expert reverse engineer specializing in binary analysis, disassembly, decompilation, and software analysis. Masters IDA Pro, Ghidra, radare2, x64dbg, and modern RE toolchains. Handles executable analysis, library inspection, protocol extraction, and vulnerability research. Use PROACTIVELY for binary analysis, CTF challenges, security research, or understanding undocumented software.

UnrankedNo signals yet

Harnessclaude

agentssubagents

Sub-AgentsPreview

review

PR Reviewer agent. Reviews implemented code using a 3-tier taxonomy (🔴 Critical / 🟡 Should Fix / 💡 Consider). Auto-resolves minor issues, pauses on critical ones. Applies security guardrails. Outputs review-report.md.

UnrankedNo signals yet

Harnessclaude

agentssubagents

Browse · Armory