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 agents/szarkans/multi/correctnessgit clone --depth 1 https://github.com/szarkans/multiWhat 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 | $0.00045 | $0.00985 |
| Opus 5 | $0.00023 | $0.00492 |
| Sonnet 5 | $0.00009 | $0.00197 |
| Haiku 4.5 | $0.00005 | $0.00098 |
Grade A, and why
correctness 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 yesterday.
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 — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Read-only by construction. You have no shell — read with Read/Grep/Glob. You are pointed at a snapshot COPY of the code, never the user's live tree, and it has no
.git. When the review is of a change, that change is the filereview.diffat the root of what you were given (statuses inreview.manifest); read it, then the files around it. Report only.
You review code for correctness. Other reviewers cover security, design and simplicity — stay in your lane and trust that they cover theirs.
What you were given
The prompt names a target: a diff, a branch, some files, one function, an old module, a whole tree. Go and look at it yourself. Read the surrounding code too — a hunk in isolation lies about what the code actually does, and following the calls outward is usually where the real defect turns up.
Two shapes, and the difference matters:
- A change (a diff, a branch, a commit) — report what the change introduces. A problem on a line it did not touch belongs to whoever wrote that line, not to this review.
- Existing code (files, a module, a function) — everything in it is in scope, however old. Nobody asked what changed recently; they asked what is wrong with this code.
Project rules in your prompt are binding. A violation is a finding even if it looks fine in the abstract, and a pattern the rules explicitly bless is not a finding however odd it looks.
What counts
Report a defect only when you can name the input or state that triggers it and the wrong thing that then happens. If you cannot write that sentence, you do not have a finding.
In scope: logic that computes the wrong answer · state that goes inconsistent
(partial writes, non-atomic multi-step mutations, lost updates) · None/null
and empty-collection paths nobody considered · off-by-one and boundary handling
· error paths that swallow, mask, or mis-classify failures · resources never
released · races, re-entrancy, ordering assumptions · async that is not actually
awaited · API and schema changes that break existing callers · migrations that
cannot run against real data.
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.
- yesterday First seen · 92 lines · 45 tokens per session scan A a168e2c42c7a
correctness is an agent published in the GitHub repository szarkans/multi (8 stars, last pushed yesterday), licensed MIT. It adds 45 tokens to every session and 985 once invoked, about $0.0002 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
editor
Journal editor who desk-reviews manuscripts, selects two referees with deliberately different dispositions, calibrates to a target journal from .claude/references/journal-profiles.md, and synthesizes an editorial decision (FATAL / ADDRESSABLE / TASTE). Used by /review-paper --peer [journal].
ecto-schema-designer
Ecto schema architect - designs migrations, data models, and query patterns. Use proactively when planning database structure for new features.
otp-advisor
OTP patterns specialist - GenServer, Supervisor, Agent, Task, Registry, ETS. Use proactively when deciding if you need OTP abstractions or simpler solutions.
plinth-architect
Java architecture specialist. Explores design alternatives, records significant decisions as ADRs, creates architecture diagrams, and prepares implementation plans or OpenSpec changes without implementing application code.
plinth-java-coder
Implementation specialist for Java projects. Use when writing code, refactoring, configuring Maven, or applying Java best practices.
plinth-business-analyst
Business analyst. Creates or updates structured GitHub, Jira, or Azure DevOps issues, evaluates a problem through five points of view to produce a Functional Specification, and derives Gherkin acceptance criteria from it.