| name | agent-builder-anthropic |
| description | Design, evaluate, and improve agentic harnesses — the orchestration layer around LLM-powered tools, agents, assistants, copilots, workflow runtimes, and AI-driven product features. Use this skill whenever the user mentions building an agentic system, structuring tool use, adding permissions or approval gates, designing multi-step AI workflows, managing context windows or memory, making agents durable or resumable, evaluating or pressure-testing an existing harness, planning phased implementation for an AI product, reviewing agent architecture, improving agent UX or observability, choosing between frameworks (LangGraph, CrewAI, Pydantic AI, smolagents, DSPy, AutoGen, DeepAgents), picking a memory substrate, or asking how to know if their harness is actually good. Also use when the user describes problems that imply harness gaps — agents doing unexpected things, context getting stale, sessions not surviving crashes, tools running without permission, or costs spiraling — even if they do not use the word "harness." This variant is optimized for Anthropic hosts (Claude Code, Claude Desktop, Claude API) and uses natural-language description for auto-activation. Also applies when building on local or open-source models (Ollama, llama.cpp, vLLM, Llama, Qwen, DeepSeek, Mistral) — see catalog/06. |
| tags | ["agentic-harness","agents","architecture","evaluation","memory","observability","frameworks","multi-agent","local-models","open-source"] |
| category | developer-tools |
| difficulty | advanced |
Agent Builder — Anthropic Variant
Use this skill as a router for designing, building, and evaluating agentic harnesses.
Read only the files you need. Do not load the entire reference set unless the request genuinely spans multiple subsystems.
Default posture:
- Bias toward lean, solo-maintainable architecture.
- Start with a single-agent design unless clear constraints justify more.
- Require an evaluation plan even for greenfield builds.
- Prefer explicit system boundaries, permission policy, and workflow state over prompt cleverness.
- Translate ideas into implementation phases, success criteria, and failure tests.
- When justifying multi-agent, cite empirical cost: single agent ≈ 4× chat tokens, multi-agent ≈ 15× chat tokens, 70%+ of multi-agent failures are systemic. See
references/catalog/01-architecture-taxonomy.md.
Step 0 — Gather Context
Before routing, make sure you have enough to work with.
For design work, confirm:
- what product or system the harness serves
- what actions the agent will take
- who the users are
- any known constraints (solo dev, existing stack, timeline, local/on-device, hardware limits)
For evaluation work, you need the harness itself:
- read the codebase, CLAUDE.md, settings, skills, hooks, architecture docs
- if evidence is missing, ask for the narrowest missing input and keep moving
- do not evaluate from vibes alone
If the request is vague ("help me build an agent" or "is my harness any good"), ask one or two clarifying questions before proceeding. Do not stall the conversation with an interview — get enough to pick a mode and start.
Step 1 — Classify The Request
Choose one mode before reading reference files.
design
User is creating a new harness, planning a major rebuild, or asking for architecture, MVP shape, or implementation sequencing.
Default reads: references/methodology/01-principles-and-solo-dev-defaults.md, references/methodology/02-harness-shapes-and-architecture.md, references/methodology/08-design-and-build-playbook.md, references/catalog/01-architecture-taxonomy.md, references/templates/design-deliverable.md. Add references/catalog/06-local-and-open-source-models.md when the target is a local/OSS model.
evaluation
User has a harness and wants gaps, risks, missing primitives, UX upgrades, or architectural cleanup.
Default reads: references/methodology/01-principles-and-solo-dev-defaults.md, references/methodology/09-evaluation-and-improvement-playbook.md, references/catalog/02-harness-components.md, references/templates/evaluation-deliverable.md.
design + evaluation
User wants a target architecture and a way to verify it, compare it with an existing system, or define acceptance criteria before building.
Default reads: union of the two above.
catalog-lookup
User is asking a factual question about what exists — "which framework", "how does Anthropic's orchestrator work", "what memory substrate", "what's the best local model setup". Route straight to the catalog. Do not dump methodology files for this mode.
Step 2 — Classify The Product Shape
| Shape | Maps to Catalog Type |
|---|
| chat assistant | Type I |
| workflow orchestrator | Type II or III |
| code agent | Type III (Claude Code / Devin / Cursor patterns) |
| internal copilot | Type I or II |
| embedded AI product feature | Type I or II |
| hybrid | Type III+ |
If the target runs on a local/open-source model regardless of shape, also read references/catalog/06-local-and-open-source-models.md.
Step 3 — Read The Smallest Useful Reference Set
Methodology
references/methodology/01-principles-and-solo-dev-defaults.md — default posture
references/methodology/02-harness-shapes-and-architecture.md — shape, boundaries, lifecycle
references/methodology/03-tools-execution-and-permissions.md — tool registries, approvals, trust tiers
references/methodology/04-state-sessions-and-durability.md — resumability, retries, idempotency
references/methodology/05-context-memory-and-evaluation.md — context, retrieval, provenance, evals
references/methodology/06-agents-and-extensibility.md — multi-agent, plugins, hooks, skills
references/methodology/07-ux-observability-and-operations.md — streaming UX, logs, budgets
references/methodology/08-design-and-build-playbook.md — build-ready plan
references/methodology/09-evaluation-and-improvement-playbook.md — findings, upgrade path
references/methodology/10-example-requests-and-output-patterns.md — prompt/output examples
references/methodology/11-codex-translation-notes.md — cross-client portability
Catalog
references/catalog/01-architecture-taxonomy.md — Type I–V, stats, debates, coordination patterns
references/catalog/02-harness-components.md — six-component harness view
references/catalog/03-frameworks.md — LangGraph, CrewAI, Pydantic AI, smolagents, DSPy, AutoGen, Bedrock
references/catalog/04-memory-substrates.md — filesystem, vector DB, in-context, COALA, Voyager, self-improvement
references/catalog/05-lab-patterns.md — Anthropic, OpenAI, Perplexity, Manus, Google, Devin, Cursor, Windsurf
references/catalog/06-local-and-open-source-models.md — Ollama / llama.cpp / vLLM, Llama / Qwen / DeepSeek / Mistral, tool-call reliability tiers, framework fit, local failure modes
Templates
references/templates/design-deliverable.md — design output shape
references/templates/evaluation-deliverable.md — evaluation output shape
Examples
examples/design-solo-pr-review-agent.md — worked design output
examples/evaluation-research-orchestrator.md — worked evaluation output
Do not rely on reference-to-reference chains. This file is the index.
Operating Rules
- Convert vague ambitions into concrete harness primitives.
- Push back on unnecessary complexity.
- Treat workflow state, permissions, context assembly, and evaluation as first-class architecture, not cleanup tasks.
- Separate universal harness primitives from product-specific manifestation.
- For evaluation requests, present findings first and improvement sequence second.
- For design requests, include how the design will be tested before calling it done.
- When recommending a framework, memory substrate, or multi-agent pattern, cite the catalog file you pulled it from.
Output Contract
For design
- recommended harness shape
- core primitives and subsystem boundaries
- MVP boundary
- phased implementation plan
- verification and acceptance criteria
For evaluation
- findings ordered by severity or leverage
- missing or weak primitives
- user experience and operational gaps
- prioritized upgrade path
- tests or checks that confirm the fixes
For design + evaluation
- target architecture
- comparison against current or likely failure modes
- implementation phases
- acceptance criteria
- evaluation plan covering regressions, safety, and UX
For catalog-lookup
- direct answer
- trade-offs relative to alternatives
- source citation (
catalog/NN-filename.md § Section)
Final Check Before Responding
- Did you keep the design lean enough for a solo developer unless the request clearly demanded more?
- Did you avoid recommending multi-agent coordination by default?
- Did you include evaluation, not just construction?
- Did you give the user an operational path forward instead of abstract theory?
- If you recommended a framework, memory substrate, or multi-agent pattern, did you cite the catalog file?
- If the target is a local model, did you apply the stricter local-model posture from
catalog/06?