agentleFS
cognitedatacognitedata/builder-skills/skills/create-agent/SKILL.md

create-agent

Skill6 starsChanged 8 days agoApache-2.0View on GitHub

MUST be used whenever scaffolding a new Atlas agent project. Creates agent.yaml, README.md, and a starter eval/cases.yaml with the correct structure, tool definitions, and instructions. Do NOT manually write agent.yaml from scratch — this skill handles the full scaffold. Triggers: create agent, new agent, scaffold agent, agent project, agent template, cognite agents create.

38 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use deploy-platform, setup-platform and 9 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • agno-agiSkill

    deploy-platform

    Deploy this AgentOS using its own provider instructions, configure JWT verification, and verify the public service. Use for first deployment, redeployment, or production configuration changes.

    Also in 5 projects that use this

  • agno-agiSkill

    setup-platform

    Set up this AgentOS from a fresh clone — confirm Docker, configure .env, boot the containers, prove the MCP endpoint live, connect the AgentOS UI, then build the user's first agent. Use when the user asks to set up the platform, get started, or bring this repo up on a new machine.

    Also in 5 projects that use this

  • ElasticSkill

    create-integration

    Use when creating a new Elastic integration package, scaffolding data streams, answering package layout or structure questions, or running the end-to-end integration build workflow. Covers package topology, scaffold commands, post-scaffold edits, and full orchestration of CEL/pipeline/test subagents.

    Also in 3 projects that use this

  • MicrosoftSkill

    create-hook

    Create a hook (.json) to enforce policy or automate agent lifecycle events.

    Also in 5 projects that use this

  • AnasX7Skill

    improve

    Survey any codebase as a senior advisor and produce prioritized, self-contained implementation plans for OTHER models/agents to execute. Strictly read-only on source code — never implements, fixes, or refactors anything itself. Use when asked to audit a codebase, find improvement opportunities (bugs, security, performance, test coverage, tech debt, migrations, DX), suggest features or where to take the project next (roadmap, product direction), or generate handoff plans for another agent to implement.

    Also in 5 projects that use this

  • ElasticSkill

    create-skill

    Interactively create a new Claude Code skill and add it to the elastic-docs-skills catalog. Use when the user wants to generate a new skill, scaffold a slash command, or build automation for docs tasks.

    Also in 13 projects that use this

  • +3 more from agno-agi

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 3 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 35 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Default instructions when the user doesn't specify:

    line 37
  2. Distinctive:

    Run the command, including only flags where the user provided a value or a non-empty default applies:

    line 49
  3. Very specific:

    Always pass --instructions (use the default if the user didn't specify).

    line 62
  4. Very specific:

    Always pass --no-prompt to skip the CLI's interactive prompts.

    line 63
  5. Very specific:

    If the user wants tools, edit <agent-name>/agent.yaml to add a tools array.

    line 70
  6. Very specific:

    Do NOT: add tools: [] — omit the key entirely when there are no tools

    line 125
  7. Distinctive:

    If a directory with the agent name already exists, warn the user and stop

    line 128

“Common” is measured: kept by at least 6 in 10 of the 35 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Create Agent

Scaffold a new Atlas agent project named $ARGUMENTS using npx @cognite/cli@latest agents create.

What gets created

<agent-name>/
  agent.yaml         # Agent definition — externalId, model, instructions, tools
  README.md          # Quick-start docs and tool reference
  eval/
    cases.yaml       # Starter eval cases (single-turn, multi-turn, correctness + faithfulness scorers)

Step 1 — Gather optional fields

The agent name is $ARGUMENTS. Ask the user for the following (skip any they already provided):

FieldCLI flagDefault
Display name--display-namesame as agent name
Description--description—
Model--modelazure/gpt-5.4-mini
Instructions (system prompt)--instructionssee default below
Tools(edit after creation)none

Default instructions when the user doesn't specify:

Help users explore and understand the data in this project.
Use the available tools to retrieve data before answering.
When the data is insufficient, say so rather than guessing.

Step 2 — Run npx @cognite/cli@latest agents create

Run the command, including only flags where the user provided a value or a non-empty default applies:

npx @cognite/cli@latest agents create <agent-name> \
  --display-name "<display-name>" \
  --description "<description>" \
  --model <model> \
  --instructions "<instructions>" \
  --no-prompt
  • Omit --display-name if the user wants the default (same as agent name).
  • Omit --description if none was given.
  • Always pass --instructions (use the default if the user didn't specify).
  • Always pass --no-prompt to skip the CLI's interactive prompts.
  • For multi-line instructions, pass them as-is in quotes; if the shell quoting is awkward, skip --instructions and edit agent.yaml manually after creation.

Step 3 — Add tools (if any)

If the user wants tools, edit <agent-name>/agent.yaml to add a tools array. Do NOT add tools: [] — omit the key entirely when there are no tools.

Available tool types

TypePurpose
analyzeDataAnalyze tabular or structured data
analyzeImageAnalyze images and P&ID diagrams
analyzeTimeSeriesAnalyze time series data
askDocumentAsk questions about documents
callFunctionCall a Cognite Function
callRestApiCall an external REST API
callWebhookPOST a payload to an external webhook
examineDataSemanticallySemantic data examination
queryStructured queries against CDF data models
queryKnowledgeGraphQuery CDF data models with natural language
queryTimeSeriesDatapointsFetch raw or aggregated time series data
runPythonCodeExecute custom Python code
summarizeDocumentSummarize documents
timeSeriesAnalysisAdvanced time series analysis and anomaly detection

Example — knowledge graph tool

tools:
  - name: find_assets
    type: queryKnowledgeGraph
    description: Find assets and related instances in the knowledge graph.
    configuration:
      version: v2
      dataModels:
        - space: cdf_cdm
          externalId: CogniteCore
          version: v1
          viewExternalIds:
            - CogniteAsset
      instanceSpaces:
        type: all

Step 4 — Verify and summarize

  1. Confirm agent.yaml, README.md, and eval/cases.yaml were all created
  2. Tell the user the next steps:
    • Edit <agent-name>/agent.yaml — add tools and refine instructions
    • npx @cognite/cli@latest agents push <externalId> — push to CDF for testing
    • npx @cognite/cli@latest agents open — open in Fusion
    • Edit <agent-name>/eval/cases.yaml and run npx @cognite/cli@latest agents eval <externalId> — the scaffolded cases are placeholders (a correctness case, a faithfulness case, and a multi-turn case); update the inputs, references, and context to match what the agent actually does

Guard rails

  • Do NOT add tools: [] — omit the key entirely when there are no tools
  • Do NOT invent tool types not listed above
  • Agent name must be kebab-case: lowercase letters, digits, and hyphens only
  • If a directory with the agent name already exists, warn the user and stop

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.