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/a777ance/claude-code-homelab/workoutgit clone --depth 1 https://github.com/a777ance/claude-code-homelabWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/a777ance/claude-code-homelab/workout)<a href="https://agentmods.dev/commands/a777ance/claude-code-homelab/workout"><img src="https://agentmods.dev/badge/commands/a777ance/claude-code-homelab/workout.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.00055 | $0.00652 |
| Opus 5 | $0.00028 | $0.00326 |
| Sonnet 5 | $0.00011 | $0.00130 |
| Haiku 4.5 | $0.00006 | $0.00065 |
Grade A, and why
workout 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 4d 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 — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Run the in-harness Jury routine on whatever the user handed you. Be forgiving about the input — a math problem, a factual call, a design decision, or open-ended prose are all fair game. Do NOT make the user reformat anything, and do NOT ask clarifying questions unless the prompt is truly unintelligible.
Prompt: $ARGUMENTS
1. Warm-up — size the question (instant)
Read the prompt and decide its shape:
- Discrete answer (a number, a name, a label, yes/no, "which of these") →
votable. Make sure the question ends with a canonical
End with 'ANSWER: <x>'.instruction (append it if it's missing) so draws tally by exact match. - Open-ended prose (explain, design, weigh trade-offs) → self-consistency voting is weaker (answers won't cluster on exact match). Still empanel a panel for the judgment, but report the result as where the panel leans, not a hard verdict — and say so.
2. Empanel the in-harness jury
Empanel 5 concurrent juror subagents (Task tool, subagent_type: "juror")
on the prompt in a single message, each with a different answer-preserving
framing (plain / skeptic / restate / cross-check / avoid-the-trap) so the draws
decorrelate by construction. Collect their ANSWER: lines, normalize, and take
the plurality. On a 3–2 / no-majority split, empanel 4 more (to 9, varying
framings) once, then stop.
For a measurable or repeatable task — where you'd want a measured jury size and a Dirichlet stopping rule rather than a fixed fan-out — the statistical jury tool (
jury_claude.py) lives in the localDNS repo under04-user-services/ai-orchestration/jury-claude/. Use it there; this command is the keyless, in-harness routine.
3. Cool-down — report honestly
Give the verdict, the tally, and a one-line confidence read (unanimous / strong majority / split). If it hit the 9-juror ceiling or split, say so plainly — a jury that won't converge means the question is genuinely contested, and that is the finding; never dress a split as a clean verdict. If the panel agrees strongly but you suspect it's confidently wrong, flag possible systematic bias — a vote can't fix that.
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.
- 4d ago First seen · 49 lines · 0 tokens per session scan A feb5f0bd96c4
workout is a command published in the GitHub repository a777ance/claude-code-homelab (2 stars, last pushed 23d ago), licensed MIT. It adds 55 tokens to every session and 652 once invoked, about $0.0003 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
git
Git operations with intelligent commit messages and workflow optimization.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.