dvy1987/agent-loom / .agents/skills/agent-builder
agent-builder
VerifiedDesign execution structure for decomposed processes: single agent or multi-agent topology. Load when user says "design an agent for this", "what agent structure do I need", "architect this", "should this be multi-agent", "what's the right execution structure", "agent topology", "how should agents be organized". Takes process-decomposer output as primary input. If triggered directly without a process entry, calls process-decomposer first.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.2 KB |
| references/examples.md | markdown | 2.8 KB |
references/examples.md
View on GitHub# Agent Builder — Full Worked Examples
Skill: `agent-builder` | Enriched from SKILL.md (improve-skills pass, SKIP_RESEARCH).
## Example 1 — Documented workflow
**Input:** Design execution for process proc-2026-04-10-007 (3 sequential steps, no parallelism)
**Output:**
```
Process: 3 steps, all sequential, no specialization.
Structure: **Single agent** with ordered skill stack.
Agent: Feature Builder
- Skills: [brainstorming, prd-writing, implementation-plan]
- Tools: [Read, Write, Bash]
- Flow: brainstorming -> prd-writing -> implementation-plan (sequential)
Generating role prompt via create-agent-prompt...
Architecture spec saved to: docs/architecture/2026-04-10-feature-builder-arch.md
```
## Example 2 — Documented workflow
**Input:** Design execution for process proc-2026-04-10-012 (5 steps, 3 parallel)
**Output:**
```
Process: 5 steps, 3 can run in parallel (code-review, security-audit, debt-check).
Structure: **Multi-agent parallel fan-out**.
Topology: Orchestrator -> [Review Agent, Security Agent, Debt Agent] -> Merge Agent
[Defining boundaries, generating role prompts for each...]
Architecture spec saved to: docs/architecture/2026-04-10-review-pipeline-arch.md
Running setup-evaluation before handoff to project-orchestrator.
```
## Example 3 — Step-by-step execution
**Input:** "Run `agent-builder` on [concrete task]"
**Agent actions:**
1. Read Process Entry
2. Decide Structure
3. Design Multi-Agent Topology
4. Write Architecture Spec
5. Hand Off
**Impact Report shape:**
```
Architecture designed for: [task]
Structure: single-agent | multi-agent-sequential | multi-agent-parallel | multi-agent-hierarchical
Agents defined: [N]
Architecture spec: docs/architecture/YYYY-MM-DD-<slug>-arch.md
Process entry linked: [proc-ID]
Next: setup-evaluation (if agent-chain) | project-orchestrator
```
## Example 4 — Gotcha application
**Input:** Task hits a non-obvious edge case
**Apply:**
- "Single agent" is not "no agent" — a single agent still needs a role prompt and skill stack.
- Platform constraints matter — check if the platform supports parallel agents before designing fan-out.
- **Scaffold choice is a first-order performance driver.** The same model varies 11+ points across scaffolds (e.g., Opus 4.6: 64.41 on Claude Code vs 53.45 on Codex). When designing architectures, consider scaffold/platform search behavior, retry strategies, and persistence characteristics — not just model capability (AlphaEval 2026, credibility 8/12).
- **When agents work on related outputs, design cross-agent validation checkpoints.** Independent optimization across agent boundaries produces globally suboptimal results — 26% cost overruns documented in procurement tasks where agents optimized components independently instead of jointly (AlphaEval 2026, credibility 8/12).
---
See `SKILL.md` for hard rules and verification checklist.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation). | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply. | Stars: 197k | Updated | Claude CodeCodex | Verified | Install |
| frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults. | Stars: 197k | Updated | Claude CodeCodex | Verified | Install |
| mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK). | Stars: 197k | Updated | Claude CodeCodex | Verified | Install |