Cognotik | ReadDocumentsTask

⚙️ ExecutionConfig.json


{
  "inquiry_goal": "Analyze the error handling strategy",
  "input_files": ["src/**/*.kt"],
  "inquiry_questions": [
    "How are network timeouts handled?"
  ]
}

→

👁️ Session UI (Analysis Tab)

Analysis: Error Handling

Based on src/network/Client.kt, timeouts are managed via a custom RetryPolicy. Exceptions are caught at the repository level and mapped to DomainResult objects...

# Files Read: 12 files matched glob pattern.

Live Results Showcase

Explore actual analysis artifacts and file context generated by this task.

Runtime Configuration

Field	Type	Description
`inquiry_goal`	String	The high-level purpose of the analysis. *
`inquiry_questions`	List<String>	Specific questions to be addressed. *
`input_files`	List<String>	Glob patterns (e.g., `*/.kt`) to include as context.
`task_description`	String	A natural language description of the analysis task.
`task_dependencies`	List<String>	IDs of tasks that must complete before this analysis runs.

* Either inquiry_goal or inquiry_questions must be provided.

Task Lifecycle

Context Resolution: Scans the workspace using provided glob patterns, filtering out ignored files.
Content Extraction: Reads text files directly; uses PaginatedDocumentReader for complex formats (PDF, DocX).
LLM Synthesis: Feeds the gathered context into a specialized "Insight" agent.
Interactive Output: Renders a tabbed UI in the session, allowing the user to ask follow-up questions.

Type Configuration

Static settings defined during task registration:

model: "gpt-4-turbo" // Default smart model

Kotlin Boilerplate

Invoke as a standalone tool using the UnifiedHarness:


import com.simiacryptus.cognotik.plan.tools.file.ReadDocumentsTaskExecutionConfigData
import com.simiacryptus.cognotik.plan.tools.file.ReadDocumentsTask.Companion.ReadDocuments

val config = ReadDocumentsTaskExecutionConfigData(
    input_files = listOf("src/**/*.kt", "README.md"),
    inquiry_questions = listOf("How is authentication handled?"),
    inquiry_goal = "Generate security documentation"
)

Prompt Segment

The following logic is injected into the orchestrator's system prompt:


ReadDocuments - Directly answer questions or provide a report using the LLM. 
Reading files is optional and can be included if relevant to the inquiry.
  * inquiry_questions: Specific questions to address.
  * inquiry_goal: The goal of the inquiry.
  * input_files: Optional file patterns to examine.