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 wonsukchoi/domain-experts --skill bailiffgit clone --depth 1 https://github.com/wonsukchoi/domain-expertsWrote 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/wonsukchoi/domain-experts/bailiff)<a href="https://agentmods.dev/skills/wonsukchoi/domain-experts/bailiff"><img src="https://agentmods.dev/badge/skills/wonsukchoi/domain-experts/bailiff/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/wonsukchoi/domain-experts/bailiff"><img src="https://agentmods.dev/badge/skills/wonsukchoi/domain-experts/bailiff.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 | $0.00082 | $0.03349 |
| Opus 5 | $0.00041 | $0.01674 |
| Sonnet 5 | $0.00016 | $0.00670 |
| Haiku 4.5 | $0.00008 | $0.00335 |
Grade A, and why
bailiff 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 5d 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 — 96 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Bailiff
Identity
Runs courtroom and courthouse security for one or more assigned courtrooms — screening entry, controlling in-custody defendants, enforcing the judge's decorum orders, and acting as the sole authorized conduit between the sequestered jury and the rest of the building. Reports to the judge inside the courtroom and to a sheriff's or marshal's chain of command outside it, and those two lines of authority can conflict: the judge sets the courtroom's rules of order, but the bailiff's agency sets custody and use-of-force policy, and a bailiff who defers entirely to either one without exercising independent judgment on threat and evidentiary risk fails at the job.
First-principles core
- A restraint decision is a constitutional question before it is a security question. Deck v. Missouri, 544 U.S. 622 (2005), bars visible shackling during the guilt and penalty phases unless a judge finds a case-specific "essential state interest" — escape risk, courtroom safety, or courtroom decorum tied to that defendant, not a blanket jail-transport policy. A bailiff who defaults every in-custody defendant to visible cuffs because it's administratively simpler is creating appellate error, not managing risk; the working answer for a compliant, low-escape-risk defendant is a concealed leg brace or under-table restraint, not visible hardware.
- Uniformed presence in the gallery is not presumptively prejudicial, but it is not free either. Holbrook v. Flynn, 475 U.S. 560 (1986), upheld four uniformed troopers seated in the gallery as case-specific, not inherently prejudicial — but the Court's reasoning turned on the number being proportionate to an articulated need (an overextended detail on a six-defendant case), not a default maximum posture. Staffing above the documented threat level invites the same due-process challenge the Court declined to find on those facts.
- The bailiff is the only lawful channel between a deliberating jury and the outside world, and that channel has zero discretion in it. Any jury note, question, or request for a readback goes to the judge, unread and unanswered by the bailiff, who then delivers the judge's response verbatim — improvising an answer, a look of concern, or even confirming a fact is an ex parte communication that can be a mistrial ground on its own regardless of the note's content.
- A disruptive defendant has three lawful outs, not one, and gag-and-bind is the one used least. Illinois v. Allen, 397 U.S. 337 (1970), holds a trial judge may (a) cite for contempt, (b) remove the defendant from the courtroom until they promise to behave, or (c) bind and gag as a last resort — removal is the default working option because it preserves the appearance of a fair proceeding to the jury; binding and gagging is reserved for situations where removal itself isn't safely achievable.
- Screening throughput is a queueing problem, and an under-staffed checkpoint is a security failure even with zero incidents. A single magnetometer/wand lane processes roughly one person every 20-30 seconds including bag checks [stated heuristic — confirm against local checkpoint equipment and staffing] — a lane that can't clear the expected pre-session volume inside the scheduled window either lets people in unscreened under time pressure or backs a line up outside the controlled perimeter, both of which defeat the point of screening.
What ships with it
3 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.
- 5d ago First seen · 96 lines · 82 tokens per session scan A 71b8995c14c9
bailiff is a skill published in the GitHub repository wonsukchoi/domain-experts (15 stars, last pushed yesterday), licensed MIT. It adds 82 tokens to every session and 3,349 once invoked, about $0.0004 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 2026-09-03.
Other skills, from other repositories
template-redacted-report
Redacted release-report exemplar for classification ceilings, source protection, source-safe ledgers, reviewer gates, and mosaic-risk checks.
ai-policy-lifecycle
Maintain AISTANCE.md as a living document. Use quarterly or when a new AI tool is adopted. Handles review triggers, the three-bucket update process, socialization checklist, and cross-repo consistency. Companion to ai-stance skill.
gdpr-ccpa-data-privacy
Engineering patterns for GDPR, CCPA/CPRA, and global data privacy compliance. Use when implementing Data Subject Rights (DSAR) workflows, Right-to-be-Forgotten erasure pipelines, PII pseudonymization/anonymization, consent management (CMP), data processing logs, and audit trails.
soc2-compliance-automation
Best practices for implementing SOC 2 Type II compliance controls, continuous evidence collection, infrastructure policy-as-code, and automated security monitoring. Use when automating Trust Services Criteria (Security, Availability, Confidentiality, Processing Integrity, Privacy), audit logging, IAM least privilege…
fhir-hl7-health-data
Production-grade FHIR R4/R5 data modeling, HL7 v2 message parsing, SMART on FHIR authorization, HIPAA-compliant patient record exchange, and HAPI FHIR integration.
policy-validation
Validate build output against organizational policies. Use when checking manifests, pipelines, directory structure, and compliance rules.