Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/obsfx/trekker-claude-code/task-agentgit clone --depth 1 https://github.com/obsfx/trekker-claude-codeWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00016 | $0.00623 |
| Opus 5 | $0.00008 | $0.00311 |
| Sonnet 5 | $0.00003 | $0.00125 |
| Haiku 4.5 | $0.00002 | $0.00062 |
Grade A, and why
trekker:task-agent scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Run as an autonomous task agent that finds ready work and completes it without user intervention.
IMPORTANT: /trekker:task-agent is a skill (invoke via Skill tool), NOT a bash command.
Mode
This is an autonomous workflow. Once started, work through tasks until:
- All ready tasks are completed
- A task requires user input/decision
- An external blocker is encountered
Workflow Cycle
1. Discovery
Find unblocked tasks ready for work:
# Check for in-progress work first (resume if exists)
trekker --toon task list --status in_progress
# Find ready tasks by priority
trekker --toon task list --status todo
Prioritize by:
- Priority (0=critical first, then 1=high, etc.)
- Skip tasks with incomplete dependencies
2. Engagement
Select a task and prepare:
# Get full task details
trekker task show <task-id>
# Check for existing comments/context
trekker comment list <task-id>
# Mark as in progress
trekker task update <task-id> -s in_progress
3. Execution
Complete the work according to the task description:
- Follow project coding standards
- Make minimal, focused changes
- Document significant decisions
4. Documentation
Track discoveries and related work:
# Create tasks for discovered bugs or improvements
trekker task create -t "New issue title" -d "description"
# Link dependencies if needed
trekker dep add <new-task-id> <original-task-id>
# Add progress comments for complex work
trekker comment add <task-id> -a "claude" -c "Progress: ..."
5. Closure
Complete the task properly:
# Add summary comment (required before completion)
trekker comment add <task-id> -a "claude" -c "Summary: implemented X in files A, B"
# Mark as completed
trekker task update <task-id> -s completed
6. Iteration
Return to Discovery phase for the next task.
Rules
- Always set status to
in_progressbefore starting work - Always add a summary comment before marking complete
- Create new tasks for discovered bugs - don't scope creep
- Stop and ask user if:
- Task requirements are unclear
- Multiple valid approaches exist
- External action is needed (e.g., API keys, permissions)
- Mark tasks
wont_fixif blocked by external factors
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 107 lines · 16 tokens per session scan A 227b20f52a18
trekker:task-agent is a command published in the GitHub repository obsfx/trekker-claude-code (4 stars, last pushed 4mo ago), licensed MIT. It adds 16 tokens to every session and 623 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
beans-release-notes
Generate release notes from commits since the last version tag and write them to .docs.local/RELEASE.md.
linear-cleanup
Audit Linear issues for problems (missing labels, stale, duplicates, wrong state, orphaned, unactionable) and propose fixes.
triage
Move tracker issues and external PRs through the triage state machine into agent-ready briefs. Use when triaging incoming issues, an external PR, or the backlog. Reads the tracker + label config written by /orc:setup.
wayfinder
Plan a chunk of work too big for one session as a shared map of decision tickets on the configured tracker; each session resolves exactly one decision. Use when an effort spans many sessions and the route isn't visible yet; /orc:plan covers single-session planning, orc:to-issues slices a settled plan.
jira-breakdown
Turn a feature brief, PRD, or existing Epic into a full Jira hierarchy behind a preview gate. Use when building out a Jira epic/backlog, breaking a feature into Jira tickets, or enforcing the Epic→Story→Task structure.
linear-bulk
Apply the same change to many Linear issues at once. Dry-run by default.