Skills

Skills inject instruction blocks into the agent’s system prompt, giving agents specialized capabilities without changing their core configuration.

List skills


GET /api/skills
Authorization: Bearer nxr_...

List builtin skills


GET /api/skills/builtin
Authorization: Bearer nxr_...

Create skill


POST /api/skills
Authorization: Bearer nxr_...
Content-Type: application/json
 
{
  "key": "security-review",
  "name": "Security Review",
  "description": "Audit code for OWASP vulnerabilities"
}

After creating, add the instruction content via the files endpoint:


PUT /api/skills/{skill_id}/files/SKILL.md
Authorization: Bearer nxr_...
Content-Type: text/plain
 
You are a security engineer. Review all code for OWASP Top 10 vulnerabilities...

Delete skill


DELETE /api/skills/{skill_id}
Authorization: Bearer nxr_...

Skill files


GET  /api/skills/{skill_id}/files
GET  /api/skills/{skill_id}/files/{file_path}
PUT  /api/skills/{skill_id}/files/{file_path}
DELETE /api/skills/{skill_id}/files/{file_path}

Export / import


GET /api/skills/{skill_id}/export
GET /api/skills/builtin/{key}/export

Built-in skills

Key	Description
`agent_spawn`	Spawn sub-agents to handle parallel workstreams
`bash`	Execute shell commands in a sandboxed environment
`git`	Local git operations
`github_read`	Read GitHub repos, issues, and PRs
`github_write`	Write to GitHub (commits, issues, PRs)
`gitlab_read`	Read GitLab projects, issues, and MRs
`gitlab_write`	Write to GitLab (commits, issues, MRs)
`platform_management`	Manage the running platform (live code/config changes)
`read_file`	Read files from the workspace
`read_url`	Fetch and read web content
`schedule_manage`	Create and manage scheduled jobs
`summarize`	Summarize long documents or conversations
`task_decompose`	Break an objective into discrete tasks
`web_search`	Search the web
`write_file`	Write files to the workspace

See the Skills UI guide for full descriptions and categories.