How this skill is triggered — by the user, by Claude, or both
Slash command
/trekker:searchThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
name: search
name: search description: PRIMARY tool for gathering context. Search tasks before any action.
Search is your PRIMARY tool for gathering context.
Before starting ANY work:
trekker search "<single keyword>"
This finds related past work using FTS5 full-text search.
Trekker uses FTS5 full-text search, NOT semantic/AI search. This means:
"fix authentication bug" requires ALL three words present, returning FEWER results| BAD (too many words) | GOOD (single keyword) |
|---|---|
trekker search "fix login authentication bug" | trekker search "authentication" |
trekker search "user cannot access account" | trekker search "account" |
trekker search "performance optimization slow" | trekker search "performance" |
trekker search "deployment pipeline issues" | trekker search "deployment" |
# Search with a single keyword
trekker search "authentication"
# Try multiple keywords separately for broader coverage
trekker search "login"
trekker search "password"
# Filter by type
trekker search "deployment" --type task
# Filter by status
trekker search "memory" --status todo,in_progress
# Rebuild index if needed
trekker search "caching" --rebuild-index
1. SEARCH: trekker search "<single keyword>"
2. BROADEN: Try 2-3 related keywords if first search has no results
3. REVIEW: Check if similar work exists
4. DECIDE: Update existing OR create new
5. SYNC: Mirror to TodoWrite after Trekker
Remember: Trekker is the source of truth.
npx claudepluginhub obsfx/trekker-claude-code --plugin trekkerLooks up tasks by ID or name, marks them in-progress, starts worklogs if enabled, executes using tools like EnterPlanMode and Bash, then completes via /complete-task. Use to pick up and execute tasks.