Generic Action Nodes

Generic action nodes provide tracker-agnostic issue operations, direct LLM inference, and text utilities. These nodes are designed for portability: issue.* nodes route through your project's active tracker binding (Linear, GitHub Issues, Jira, or Asana) without requiring tracker-specific configuration on the node itself.

Issue nodes (tracker-agnostic)

The issue.* family projects operations onto a canonical issue shape and routes to whichever tracker is configured as the project's primary tracker. This makes workflows portable across organizations using different trackers.

issue.read

Reads an issue from the project tracker and returns it in the canonical cross-provider shape.

Input ports

Output ports

IssueReadResult shape

{
  ok: boolean
  issue?: {
    id: string
    identifier: string    // e.g. "REN-1234" or "#42"
    title: string
    description?: string
    status: string
    url: string
    labels?: string[]
    assigneeId?: string
    provider: string      // "linear" | "github_issues" | "jira" | "asana"
  }
  details?: object        // Raw provider response for debugging
  reason?: string         // Error message when ok=false
}

issue.update_status

Updates the status of an issue using its canonical tracker state name. The node handles the translation from the canonical name to the tracker-native state identifier.

Input ports

Output ports - success ({ updated: boolean }) / fail ({ error: string })

issue.add_comment

Adds a comment to an issue on the project's active tracker.

Input ports

Output ports - success ({ id: string }) / fail ({ error: string })

LLM nodes

The llm.* pair separates configuration from execution. Declare a shared llm.config node and wire it to one or more llm.inference nodes - this lets you swap models or adjust parameters in one place rather than on every inference node.

llm.config

Declares a reusable LLM configuration as a first-class graph node. Its output port carries an LlmConfig object consumed by downstream llm.inference nodes.

Config schema

Output port - output (LlmConfig): the configuration object, ready to wire to llm.inference.

llm.inference

Calls an LLM and returns the generated text (and optionally parsed JSON). Accepts either a prompt string shortcut or a full messages array, plus optional overrides for model, temperature, max tokens, and system prompt.

Input ports

Config schema (inline fields, supplement or override config)

Output port - output (InferenceResult)

InferenceResult shape

{
  text: string             // Generated text
  json?: object            // Parsed JSON when jsonMode=true
  finishReason?: string    // "stop", "length", "tool_calls", etc.
  usage?: {
    promptTokens: number
    completionTokens: number
    totalTokens: number
  }
  model?: string           // Actual model used
  latencyMs?: number       // End-to-end latency
}

Example: classify an issue

steps:
  - id: llm_cfg
    type: action
    nodeId: llm.config
    config:
      provider: claude
      model: claude-opus-4-8
      temperature: 0
      maxTokens: 256
      systemPrompt: "Classify the following issue as bug, feature, or chore. Respond with JSON."

  - id: classify
    type: action
    nodeId: llm.inference
    config:
      prompt: "{{ nodes.readIssue.output.issue.title }}: {{ nodes.readIssue.output.issue.description }}"
      jsonMode: true

  - id: check_class
    type: condition
    nodeId: rensei.value.switch
    config:
      value: "{{ nodes.classify.output.json.classification }}"
      cases:
        - bug
        - feature
        - chore

Text utility

rensei.text.keyword_prefix

Prepends a keyword prefix to a string value. Used in the SDLC templates to stamp work-type labels (e.g. "Research: ", "Backlog: ") onto comment bodies or issue titles before passing them downstream.

Input ports

Output port - output ({ text: string }): "${prefix}${text}"

When to use issue.* vs provider-specific nodes

Related pages

• Linear Action Nodes - full Linear issue lifecycle
• GitHub Issues Action Nodes - full GitHub Issues lifecycle
• Condition Nodes - rensei.agent_result.switch, rensei.value.switch
• Foundational Nodes - llm-model - project an LLM profile ref onto llm.config