claude-agents-creator
Create or revise Claude Code subagents or Claude Managed Agents, including definitions, delegation, permissions, and API sessions.
v2026.11.13-21cd5683 instalaciones+3 en 7 días39 EstrellasMITsincronizado hace 11 horas
Instalar
¿Aún no tienes la CLI de NotHoly? Instálala primero.
$ curl -fsSL https://notholy.com/install.sh | sh$ notholy skills install aeondave/claude-agents-creatorEscribe en ~/.notholy/skills después de comprobar la firma y el SHA-256.
FirmaFirmado, ed25519
Paquete75f4c6f1…02b2f746
75f4c6f149f7…ba6602b2f74610 archivos · 11.8 KBEscaneoMedio2 hallazgos
- injection.no_confirmationreferences/frontmatter-reference.md:11
- Enlaces a dominios externos
Fuentegithub.com/AeonDave/malskill/tree/main/harness-dev/claude-agents-creator@21cd568
Esta entrada todavía no está escrita en tu idioma, así que aparece en English.
En esta página
Create a focused worker with a concrete trigger, capability boundary, and usable result. Default to a project-scoped Claude Code subagent for repository work; use Managed Agents when an application needs the hosted API harness.
For repository instructions use agent-md-creator; for reusable Agent Skills use skill-creator; for OpenCode workers use opencode-agent-creator.
Build the definition
- Establish the host, installed version, scope, task, and completion signal. Use an existing built-in worker when it already fits. Derive routine choices from the request; ask only for missing information that changes the result.
- Give the worker one responsibility. Put the concrete delegation trigger in a short
description; put workflow and output requirements in the body. - Choose project
.claude/agents/, personal~/.claude/agents/, or pluginagents/. Keep identities unique; check collision precedence and plugin restrictions in references/frontmatter-reference.md. - Set the tools the work requires. For a reviewer that must not write, start with
Read, Grep, Glob; shell and MCP access can permit writes, and persistent memory can add write tools. A prose boundary alone does not restrict capabilities. - Choose an available model alias or ID. Set
model: inheritwhen matching the parent is intended; an omitted field can resolve through other model settings. Add skills, memory, MCP, hooks, or isolation only when the task needs them. - Write the body: role, required inputs, essential steps, completion criteria, and returned evidence. Keep stable conventions in the body or preloaded skills; pass the current task's files, constraints, and observations in the delegation message.
---
name: change-reviewer
description: Reviews a supplied change set and reports defects with file references. Use after implementation changes.
tools: Read, Grep, Glob
model: inherit
---
Review the supplied diff and affected files. Report defects with file:line,
impact, and a concrete correction. If no diff was supplied, name the missing
input and inspect the target files without invoking a shell. Do not edit files.Context and delegation
A named subagent normally starts without the parent's conversation or previously read files. Preloaded skills: supply full skill content. An explicit conversation fork inherits the conversation; a skill's context: fork still starts a separate subagent without that history. Load the frontmatter reference when choosing between these modes, nested delegation, or worktree isolation.
Verify before handing back
- Check YAML, required identity/description, supported field spelling, available tools/model, and the intended discovery scope.
- Parse frontmatter with a strict YAML parser and check required fields. Run
claude plugin validate <path-to-agents-dir>for additional component lint; use a directory namedagents. Its success does not establish valid YAML, routing, or effective permissions. - Run a representative task and a plausible near miss; inspect the actual model, tool calls, result, and any partial completion. Name runtime checks that were not executed.
Load references/triggering-and-testing.md when checking discovery, delegation, capability boundaries, or reload behavior.
Load when needed
- references/frontmatter-reference.md: load when setting fields, permissions, model precedence, context, MCP, hooks, or CLI definitions.
- references/system-prompt-patterns.md: load when designing the body, delegation brief, or coordinator hand-off.
- references/managed-agents.md: load for hosted API agents, versioned sessions, events, or environment configuration.
- assets/agent-template.md: adapt when creating a Markdown definition.
- assets/examples/code-reviewer.md, debugger.md, safe-researcher.md, test-runner.md: load the example matching the requested worker.
Archivos de esta versión
assets/56.0 KB
examples/44.7 KB
- code-reviewer.md1.4 KB
- debugger.md1.2 KB
- safe-researcher.md1.1 KB
- test-runner.md1.1 KB
- agent-template.md1.3 KB
references/419.6 KB
- frontmatter-reference.md9.1 KB
- managed-agents.md3.8 KB
- system-prompt-patterns.md3.5 KB
- triggering-and-testing.md3.2 KB
- SKILL.md4.6 KB
Permisos
| Permiso | Declarado | Hallado en el escaneo |
|---|---|---|
| Lee archivos | Sí | Sí |
| Escribe archivos | Sí | Sí |
| Ejecuta comandos de shell | Sí | Sí |
| Usa la red | Sí | Sí |
| Usa conectores | Sí | Sí |
| Controla un navegador | Sí | Sí |
Hallazgos del escaneo
Revisado el 13 nov 2026
- MediaPrompt injection · injection.no_confirmationreferences/frontmatter-reference.md:11
- InfoEnlace · Enlaces a dominios externos
Un escaneo limpio baja el riesgo, pero no prueba que el paquete sea seguro. Lee lo que instalas.
| Versión | Publicada | Commit | SHA-256 | Firma | Cambio de permisos |
|---|---|---|---|---|---|
| v2026.11.13-21cd568 | 13 nov 2026 | 21cd568 | 75f4c6f1 | Firmada | – |
Más de AeonDave
Todo de este publicadorMás en APIs, SaaS e identidad
Ver la categoría- testing-for-sensitive-data-exposuremukul975/testing-for-sensitive-data-exposure5
- performing-jwt-none-algorithm-attackmukul975/performing-jwt-none-algorithm-attack5
- sharp-edgestrailofbits/sharp-edges4
- hydraaeondave/hydra4
- implementing-google-workspace-admin-securitymukul975/implementing-google-workspace-admin-security4
- implementing-api-security-posture-managementmukul975/implementing-api-security-posture-management4
Pregunta todo lo que tu alcance permite.
Crea la cuenta, añade créditos y haz la primera pregunta en el chat o en la terminal.
- 1/3
Crea la cuenta
Individual o empresa. Puedes pasar a empresa después sin perder saldo ni claves.
IndividualEmpresa - 2/3
Añade créditos
Con tarjeta. Pagas por token, sin suscripción.
$5.00recarga mínima - 3/3
Pregunta
En el chat, o instala el CLI y ejecuta notholy login en tu terminal.
$ notholy login