external-agents

Connect external agents and MCP hosts (Claude, Claude Desktop, Claude Code, ChatGPT custom MCP apps, Codex, Cursor, Claude Cowork, VS Code GitHub Copilot, Goose, Postman, MCPJam) to an agent-native app over MCP, and round-trip artifacts back into the UI with MCP Apps and deep links. Use when adding an action's `link` builder or `mcpApp`, wiring the `/_agent-native/open` route, exposing an "ingest" action to MCP/A2A, or scaffolding apps from an external agent.

Quality

83%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

75%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a tight, actionable dev guide with concrete code, commands, real file paths, a clear numbered workflow, and validation guidance, all supported by a well-signaled reference file. It is consistently strong but leaves minor room to tighten prose, complete a few elided examples, and externalize more detail.

Suggestions

Tighten the longer prose passages (e.g., the MCP 2026-07-28 protocol-version paragraph and the reconnect subcommand narration) into terse rules to push conciseness toward level 5.

Complete the elided action examples (fill in schema/run for the `embedApp()` and `manage-draft` blocks) so the main code samples are fully copy-paste ready.

Consider moving the protocol-versioning and auth/authz detail blocks into a reference file so the main body stays an overview pointing one level deep.

Dimension	Reasoning	Score
Conciseness	The body assumes Claude's competence (no explaining of MCP/OAuth/deep-links) and uses a terse Rule/Why/How/Do/Don't structure, but at ~530 lines a few prose-heavy passages (the MCP 2026-07-28 protocol paragraph, the reconnect subcommand narration) could be trimmed, matching the efficient-with-minor-over-explanation level-4 anchor rather than the every-token-earns-its-place level 5.	4 / 5
Actionability	Provides concrete executable code (the `manage-draft` `link` builder, `embedApp()`, `callMcpTool`), specific commands (`npx @agent-native/core@latest connect ...`, `mcp install ...`), and real file paths, but some examples elide schema/run with `// ...schema, run...`, leaving minor gaps that fit the mostly-executable level-4 anchor.	4 / 5
Workflow Clarity	The numbered How section (1, 1a, 1b, 2, 2a, 3, 4, 5) gives a clear sequence with validation guidance ("validate with a fresh tool call", "test real ChatGPT/Claude web behavior with a fresh inline render", sparse-tools/list diagnostics) and Do/Don't checklists, but the validate→fix→retry feedback loops are not fully explicit, fitting level 4 rather than the explicit-recovery-loops level 5.	4 / 5
Progressive Disclosure	Clean section structure with a single clearly-signaled, verified one-level-deep reference (`references/mcp-apps-embedding.md`, "Read it before changing the `embedApp()` shell") plus pointers to related skills and source files; content is mostly appropriately split, though substantial detail (protocol versioning, auth/authz) remains inline, fitting level 4 rather than the cleanly-split level 5.	4 / 5
	Total	16 / 20 Passed

Description

92%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is concrete, complete, and highly distinctive, pairing a clear statement of capability with an explicit "Use when" trigger clause. It is strong throughout, with only minor room to broaden natural-language trigger phrasing.

Dimension	Reasoning	Score
Specificity	Names the domain and lists multiple concrete actions — "round-trip artifacts back into the UI with MCP Apps and deep links", "adding an action's `link` builder or `mcpApp`", "wiring the `/_agent-native/open` route", "exposing an 'ingest' action", "scaffolding apps" — with comprehensive coverage of the skill's scope, matching the score-5 anchor rather than the minor-gaps level 4.	5 / 5
Completeness	Clearly states what it does (connect external agents/MCP hosts over MCP and round-trip artifacts via MCP Apps and deep links) and gives an explicit "Use when..." clause with concrete trigger phrases, exactly matching the score-5 anchor for both what and when.	5 / 5
Trigger Term Quality	The "Use when" clause and host enumeration (Claude, ChatGPT, Cursor, Codex, etc.) give good keyword coverage of natural terms, but several triggers are route-level implementation details ("wiring the `/_agent-native/open` route", "ingest action to MCP/A2A") rather than user-facing phrases, leaving a few natural variations missing — the level-4 anchor.	4 / 5
Distinctiveness Conflict Risk	The niche is sharply defined (external-agent MCP bridging to an agent-native app, deep links, MCP Apps) with a specific enumerated host set, giving it clear distinct triggers and minimal conflict risk with other skills.	5 / 5
	Total	19 / 20 Passed

Validation

68%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 11 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (549 lines); consider splitting into references/ and linking	Warning
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 1 missing	Warning

	Total	11 / 16 Passed

Repository: BuilderIO/agent-native
Path: .agents/skills/external-agents/SKILL.md
Commit: a06e0ac

Reviewed: 13 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.