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 skills add auerbachb/claude-code-config --skill skillgit clone --depth 1 https://github.com/auerbachb/claude-code-configWrote 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/skills/auerbachb/claude-code-config/skill)<a href="https://agentmods.dev/skills/auerbachb/claude-code-config/skill"><img src="https://agentmods.dev/badge/skills/auerbachb/claude-code-config/skill/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/auerbachb/claude-code-config/skill"><img src="https://agentmods.dev/badge/skills/auerbachb/claude-code-config/skill.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00064 | $0.05441 |
| Opus 5.5 | $0.00026 | $0.02176 |
| Sonnet 5.5 | $0.00013 | $0.01088 |
| Haiku 4.5 | $0.00006 | $0.00544 |
Grade A, and why
desk 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 7d 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 — 165 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/desk — the human queue's control session
The desk is the operator's single point of contact (desk/DESIGN.md 2.3, 2.6). Worker threads' questions are captured into the store by the capture hook (desk/README.md, "Capture hook") and render only here, as the same clickable menus a normal session shows. Answers go to the store; the asking thread gets a pointer and reads the store.
The hook queues a question only while a desk is live: a registered control session whose last tick is at most 15 minutes old. Running /desk is what makes it live, so the steps below are exact — a desk that believes it is ticking and is not leaves every worker's questions with nobody to show them to.
This file is a router. It owns starting the desk, the Monitor, its events, and the end-of-turn gate. Each kind of work lives in its own file, loaded when it is needed:
| File | Owns | Issue |
|---|---|---|
decisions.md |
Simple Decisions: sets, menus, replies, answers, wake-ups | #1779 |
longform.md |
Long-form and multipart Decisions: one text prompt at a time, part by part, answers stored word for word | #1780 |
discuss.md |
discuss <n|D-id>: talk one item through with its context loaded, then answer it |
#1780 |
desk.jq |
The functions both views call: which Decisions fit a menu, sets, multipart groups, the long-form and discussion cards, feedback tags | #1779, #1780, #1783 |
interrupts.md |
desk/policy.json; the interrupt rule (away, available, focus until …, focus for …, focus off, interrupts?); feedback tags (2: not important, 2: should have defaulted, 2: good interrupt) |
#1783 |
wakeups.md |
Wake-up retries on the next three ticks, then answer-parked, shown once |
#1781 |
history.md |
show D-<n> (an item's sub-thread) and history (today's answered items), printed without a state line |
#1781 |
priorities.md |
top, bump, park, drop, priorities: the operator's backlog order for /pm, kept in the target repo's .claude/pm-priority.json |
#1767 |
reviews.md |
The Reviews view: reviews (one line per unreviewed item, by day and repo), open R-<n> (level 2, cached), diff R-<n> [path] (level 3, live), reviewed, reviewed all today, flag, and follow up |
#1782 |
drilldown.md |
The PR drill-down: outline R-<n> (files, hunks, and the tests touching each, numbered), open R-<n> <node>… (a hunk with twenty lines of context, or a file's diff), ask R-<n>: <question> (answered from the hunks it reads, named on a Read: line); nothing stored |
#1768 |
ideas.md |
idea: <text> / file: <text>: file an issue through /issue-maker's one-shot entry, no capture mode; repo: owner/name sets where ideas go |
#1766 |
plan.md |
The day plan: plan, I need to work on …, plan: … (one sentence revises it), plan?, plan off; asks pace and chunking, forecasts incoming questions, proposes what to clear first, stores the confirmed blocks, which hold new Decisions while they run |
#1784 |
sweep.md |
The end-of-day sweep at eod_time (the eod event) or on sweep: everything still open as one numbered list, then the paper export (export, #1759) |
#1784 |
attention.md |
The weekly attention report: report (report <YYYY-MM-DD> for another week), one page from the events table, offered once a week after Friday's end-of-day sweep |
#1771 |
todo.md |
The operator's own to-do layer on items: tag, untag, note, unnote, snooze … until|for …, unsnooze, mine D-<n> 1-5|off, and my list (by personal priority, then age; snoozed items hidden until their time). Not /pm's priorities |
#1769 |
checkin.md |
The morning check-in (the morning event, check-in, or a new plan with none today): hours at the desk, energy, plans, and yesterday's measured pace set today's reading budget; budget? |
#1770 |
What ships with it
17 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 7d ago Changed · +16 lines 500b4cb15dd5
- 8d ago First seen · 149 lines · 64 tokens per session scan A 1e722d45c2df
desk is a skill published in the GitHub repository auerbachb/claude-code-config (5 stars, last pushed yesterday), licensed MIT. It adds 64 tokens to every session and 5,441 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-10-08.
Other skills, from other repositories
Review Comment Triage (No-Code-Fix Mode)
A review workflow for sorting existing pull-request comments by severity, status, and next action without changing the code.
recipe-front-review
Reviews completed frontend implementation for governing-source compliance, scope economy, repository quality, and security, then applies user-approved React corrections.
pr-description-writer
Redacta una descripción de pull request clara y estructurada a partir de un diff de git, resumen de rama o lista de commits. Úsalo cuando te pidan escribir una descripción de PR, redactar una solicitud de cambios o documentar cambios de código. Produce una descripción con resumen, motivación, cambios realizados, pasos…
output-eval-error-analysis
Systematically review workflow traces to identify failure modes before building evaluators. Use when starting an eval project, after significant pipeline changes, or when production quality drops.
rag-code-review
When performing code review on a PR, diff, snippet, or "look at this change" request, first consult the corpus for related ADRs, coding standards, prior patterns, and similar files. Grounds review comments in the team's actual decisions instead of generic best practices. Trigger on any review-style request — "review"…
idea-team
A structured group discussion for exploring a product idea through three roles: researcher, critic, and analogy finder. The roles look for evidence, weaknesses, and useful comparisons.