Sequential Thinking

• thought (string): Content of your thought/research step.
• thoughtNumber (integer): Current step number (e.g., 1).
• totalThoughts (integer): Expected total number of analysis steps.
• nextThoughtNeeded (boolean): true until you reach the synthesis stage.

Optional parameters:

• isRevision (boolean): Whether you are correcting a previous thought.
• revisesThought (integer): Number of the thought you are correcting.
• branchFromThought (integer): Number of the thought from which you create a new reasoning branch.
• branchId (string): Branch identifier.
• needsMoreThoughts (boolean): Whether you need more steps than originally planned.

Compatibility with open-websearch

If the workflow uses the open-websearch skill, the thought plan should explicitly include:

• the main topic and a list of 2–5 working queries,
• a decision whether to use a single fetchWebContent or the collect-web-research.mjs script,
• the target main directory in research_data/<research-folder>/ and artifacts subdirectory zrodla_i_analiza/,
• criteria for reviewing source-index.md and saved .md files before synthesis,
• how to handle missing or contradictory sources.

Recommended thought sequence

1. Understanding the topic and creating structure for research_plan.md.
2. Breaking the topic into research questions and search queries.
3. Deciding whether source collection will be done manually or via open-websearch/scripts/collect-web-research.mjs.
4. Defining source selection criteria and quality risks.
5. Reviewing local results in research_data/<research-folder>/zrodla_i_analiza/, including source-index.md and saved .md files.
6. Synthesizing conclusions and preparing the final report.

Best practices

• The first thought should always be: "Understanding requirements and creating structure for research_plan.md".
• If using open-websearch for multiple queries, one of the early thoughts must explicitly list the queries, the main research directory, and the target directory research_data/<research-folder>/zrodla_i_analiza/.
• Before drawing conclusions for the final report, your second-to-last thought must contain a review of fetched sources from research_data/<research-folder>/zrodla_i_analiza/, with special attention to source-index.md and local .md files.
• If some sources were not fetched, use thought revision or branching to decide whether to narrow the topic, change queries, or continue analysis based on already locally saved sources.