| name | spec_navigator |
| description | Efficiently retrieve specific sections from requirement documents (prompts) to save context tokens. |
Spec Navigator Skill
This skill allows you to query specific sections of the master instruction file (inst.md) or any markdown file, saving you from reading the entire file into the context window of the LLM.
Purpose
To significantly reduce token consumption when working with large specification files like inst.md by only loading the relevant section (e.g., "Architecture", "Stack", "MVP") into the conversation context.
Available Tools
scripts/query_md_section.py
Extracts a section and its subsections from a markdown file based on a header search term.
Usage:
python3 .agent/skills/spec_navigator/scripts/query_md_section.py [ABSOLUTE_PATH_TO_FILE] [HEADER_SEARCH_TERM]
Workflow Example:
- User asks: "What is the database configuration?"
- Instead of reading the whole
inst.md, run:
python3 .agent/skills/spec_navigator/scripts/query_md_section.py /home/crrb/ag_projects/portal_da-tica/inst.md "Persistencia"
- The output will be just the ~10 lines related to persistence, saving ~150 lines of context.
Common Search Terms for inst.md
"Arquitectura" - For stack and module layout."MVP" or "Alcance" - For the immediate to-do list."Persistencia" - For database and volume mounts."Restricciones" - For negative constraints (what NOT to do).
