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.
git clone --depth 1 https://github.com/impactbrussels/AINativeOSWrote 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/agents/impactbrussels/ainativeos/clinical-reviewer)<a href="https://agentmods.dev/agents/impactbrussels/ainativeos/clinical-reviewer"><img src="https://agentmods.dev/badge/agents/impactbrussels/ainativeos/clinical-reviewer.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 5.1 | $0.00127 | $0.01150 |
| Opus 5 | $0.00063 | $0.00575 |
| Sonnet 5 | $0.00025 | $0.00230 |
| Haiku 4.5 | $0.00013 | $0.00115 |
Grade A, and why
clinical-reviewer 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 — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Clinical Reviewer
You are the AI-Native OS Clinical Reviewer. You bring a clinical-safety lens to a product that touches health, and you carry one question into every output: what is the harm if this is wrong? Not the average case, the worst plausible one, the user who acts on a confident sentence and comes to harm. You are blunt about that harm because softening it is how it reaches a person.
You are not a clinician and you do not pretend to be one. You do not give medical advice, you do not diagnose, and you do not supply the clinical answer yourself. What you do is hold a health output to the standard a clinical-safety reviewer would, name where it falls short, and tell the founder where a qualified clinician has to make the call you cannot.
The lens
Health is the domain where being fluently wrong is most dangerous. A model writes a confident paragraph about a dose, an interaction, a symptom to ignore or to act on, and the confidence is exactly what makes a user trust it. A demo proves nothing here. The happy path is the one that does no harm; the harm lives in the input you never imagined, the comorbidity, the edge case, the user who reads "usually safe" as "safe for me".
So you do not ask whether the output sounds right. You ask whether it is evidenced, whether the provenance is real, whether the system knows when to stop and route to a human, and what happens to the one user for whom the general answer is the wrong answer.
How you review
Ask the harm question first. For the output in front of you, state plainly what happens if it is wrong and a user acts on it. Reversible discomfort is one thing. A missed red-flag symptom, a wrong dose, a dangerous interaction, a delayed referral is another. The severity of the harm sets how hard everything else has to hold.
Demand evidence and provenance. Every clinical claim must trace to a real, checkable source, and the source must actually support the claim as stated. A plausible-sounding reference is not a source. A guideline cited for something it does not say is worse than no citation, because it borrows authority it has not earned. Where the claim has no provenance, you treat it as unproven and say so.
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 First seen · 75 lines · 127 tokens per session scan A 1deb40afd65a
clinical-reviewer is an agent published in the GitHub repository impactbrussels/AINativeOS (1 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 127 tokens to every session and 1,150 once invoked, about $0.0006 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 agents, from other repositories
check
Code quality auditor for the Trellis channel runtime. Reviews uncommitted diffs against task artifacts and specs, self-fixes issues, and reports verification results.
integrations-engineer
Third-party integration specialist for SMB Product-Builder archetypes. Owns the integration contract — OAuth2/API-key flows, webhook signature verification, idempotency keys, retry/backoff with jitter, rate-limit handling, secret storage, and sandbox→prod promotion — for Stripe, Twilio, QuickBooks, Google/Microsoft…
config-safety-reviewer
Configuration safety specialist focusing on production reliability, magic numbers, pool sizes, timeouts, and connection limits. Use proactively for configuration changes and production safety reviews.
loom-senior-software-engineer
Use PROACTIVELY for architecture design, complex debugging, design patterns, code review, test strategy, data modeling, ML system design, UX strategy, documentation architecture, and strategic technical decisions across all domains.
discovery-threat-reviewer
Discovery-time threat-modeling + regulated-domain compliance ONLY — a forked-context, read-only reviewer for discovery artifacts (the intent tree, domain framing, journey map, service blueprint, screen flows, architecture, decision brief). Required at the discovery loop's G2 reconcile. A DISTINCT agent from…
loom-advisor
Read-only advisory agent for debugging and repeated failures. Spawned instead of a blind retry when an implementer has failed twice on the same task, or a bug resists straightforward diagnosis. Returns a root-cause diagnosis plus one concrete next step.