The Agent Skills Directory

[SAFE]: The skill provides a systematic reasoning methodology for complex problem-solving. The instructions and scripts are designed to facilitate structured analysis, hypothesis verification, and context management without engaging in restricted or dangerous behaviors.
[INDIRECT_PROMPT_INJECTION]: The skill processes 'thoughts' that may contain data from untrusted external sources if the agent is used to analyze such content. This constitutes a potential injection surface, though the scripts themselves do not execute the content.
Ingestion points: Reasoning content is ingested via the --thought argument in scripts/process-thought.js and scripts/format-thought.js.
Boundary markers: The methodology uses structural markers (e.g., 'Thought 1/5') to organize reasoning, but no specific sanitization or delimiters are applied to the internal thought text.
Capability inventory: The script scripts/process-thought.js performs local file operations (write and delete) for maintaining a reasoning history file named .thought-history.json within the skill directory.
Sanitization: The scripts validate input types (e.g., ensuring thoughts are strings and numbers are positive) but do not filter for natural language instructions within the thought strings.

sequential-thinking