create-agent
Create and update Harness AI agent instances - standalone templates for agentic workflows in pipelines. Use when asked to create agent, update agent, modify agent spec, build autonomous systems, or work with AI agents.
88
85%
Does it follow best practices?
Impact
89%
6.84xAverage score across 3 eval scenarios
Low
Low-risk findings worth noting
Evaluation results
84%
71%Code Coverage Enforcer Agent
Agent spec with custom inputs — defaults and expression syntax
UID prefix and slugification
0%
0%
Default for repo_name input
0%
100%
Default for branch input
0%
100%
Default for threshold input
0%
100%
Expression syntax in env vars
0%
100%
allowedDomains input present
0%
0%
PLUGIN_ALLOWED_DOMAINS env var
0%
100%
allowedDomains in layout
0%
100%
Custom inputs absent from layout
100%
100%
PLUGIN_MAX_TURNS range
0%
100%
Required env vars present
0%
100%
Correct container image
0%
100%
No clone/platform sections
100%
100%
RULES section in PLUGIN_TASK
0%
100%
85%
75%Database Migration Checker Agent
Harness AI agent spec structure and UID
Correct UID value
0%
0%
UID explicit in spec
0%
0%
step.group.steps structure
0%
100%
Step name field
0%
100%
Step if field
0%
100%
Step id field
0%
100%
Exact container image
0%
100%
PLUGIN_TASK present
0%
100%
RULES section in PLUGIN_TASK
0%
100%
PLUGIN_MAX_TURNS present
0%
100%
PLUGIN_HARNESS_CONNECTOR
0%
100%
PLUGIN_ALLOWED_DOMAINS
0%
100%
llmConnector input
0%
100%
allowedDomains input
0%
100%
Layout block contents
0%
100%
No forbidden sections
100%
100%
No modelName/ANTHROPIC_MODEL
100%
100%
100%
82%PR Review Agent
Agent spec with MCP connector for external service integration
PLUGIN_MCP_FORMAT env var
0%
100%
PLUGIN_MCP_SERVERS exact syntax
0%
100%
mcpConnectors type: array
0%
100%
mcpConnectors ui.component: array
0%
100%
mcpConnectors connector inputType
0%
100%
mcpConnectors connectorTypes: [Mcp]
0%
100%
mcpConnectors in layout
0%
100%
No modelName or ANTHROPIC_MODEL
100%
100%
allowedDomains input present
0%
100%
PLUGIN_ALLOWED_DOMAINS env var
0%
100%
allowedDomains in layout
0%
100%
Connector placeholder IDs
100%
100%
- Repository
- harness/harness-ai
- Commit
e554080
- Evaluated
- Agent
- Claude Code
- Model
- Claude Sonnet 4.6
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.