Codex skill for using Trekker as persistent task and issue memory across sessions.
This repository contains one skill:
trekker- a Codex-native workflow for search-first task tracking, context recovery, checkpoint comments, and completion summaries in Trekker
This skill does not install Trekker itself. It teaches Codex how to use an existing Trekker setup correctly.
Before installing the skill, make sure you have:
- Codex
- Trekker CLI installed
- a project with Trekker initialized, or permission to run
trekker init
Install Trekker CLI:
npm install -g @obsfx/trekkerInitialize Trekker in a project:
trekker initInstall the skill with the Skills CLI:
npx skills add https://github.com/obsfx/trekker-skill --skill trekkerAfter installing, restart Codex so it picks up the new skill.
Open a project that already uses Trekker, then ask Codex to do something that should trigger the skill. For example:
continue the current Trekker taskuse Trekker and show me what is readycheck existing Trekker tasks before creating a new one
The skill should auto-activate when the request matches Trekker usage. When it does, Codex should prefer Trekker as the source of truth and use direct trekker CLI commands.
When active, the skill guides Codex to:
- use Trekker as the persistent source of truth instead of session-local planning state
- search before creating, starting, or revisiting work
- restore context with
trekker search,trekker task show,trekker comment list, andtrekker history - add checkpoint comments during work that spans sessions
- add a
Summary:comment before marking a task complete - avoid Claude Code-only concepts such as slash commands, hooks, and MCP tools