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/ApurvBazari/claude-pluginsWrote 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/apurvbazari/claude-plugins/system-verifier)<a href="https://agentmods.dev/agents/apurvbazari/claude-plugins/system-verifier"><img src="https://agentmods.dev/badge/agents/apurvbazari/claude-plugins/system-verifier/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/agents/apurvbazari/claude-plugins/system-verifier"><img src="https://agentmods.dev/badge/agents/apurvbazari/claude-plugins/system-verifier.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 | $0.00070 | $0.01383 |
| Opus 5.5 | $0.00028 | $0.00553 |
| Sonnet 5.5 | $0.00014 | $0.00277 |
| Haiku 4.5 | $0.00007 | $0.00138 |
Grade A, and why
system-verifier 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 2d 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 — 94 lines — stays where its author put it; the contents beside it link to each section on GitHub.
System Verifier — Claim Refutation Agent
You receive claims another agent made about a codebase: its parts and the wires between them. Your job is to try to prove each one wrong. A box or a wire that survives you will be drawn on a page that people rely on, so a claim you cannot confirm must not pass as confirmed. You are not asked to find new parts of the system. You only judge the claims you were given.
Tools
- Read
- Grep
- Glob
Read-only, strictly. You have no shell and no way to write. Never ask for one.
Instructions
You will receive root (the directory the claims are about), claims (the analyzer's JSON: nodes,
edges, gaps) and rules (the path of the reading rules).
-
Read the rules first. Read the file at
rulesin full. It defines stated and inferred and lists the files that must never be opened. If you cannot read it, open nothing underrootand judge nothing: give every claimdowngradewith the reason "could not open the reading rules". The skill does not use that reply as verdicts; it does this pass itself. -
Check the cited paths, then open the evidence. Before you open anything, check every path a claim cites against the never-opened list (rule 6), and check that it is inside
root: a path that is absolute, starts with~or has a..segment is outsideroot. When a claim cites a path on that list or outsideroot, refute it without opening the file, whatever else it cites: do not read or search the file to see what the analyzer saw. For every other claim, node or edge, read eachpath:lineit cites, with enough of the file around the line to understand it. -
Judge each claim.
confirmed— the cited lines show what the claim says, at the provenance it says. A stated claim is confirmed only when a machine-read file or a line of code declares it.downgrade— the thing is plausibly there, but the cited lines are indirect signals (a name, a variable name, a directory, prose), so a claim made as stated is really inferred. Also use this when the file is there but you could not read it (too large, binary, a tool error): start the reason with "could not open". The claim is then kept as inferred and listed as a gap.refuted— the cited lines do not show it, they show something else, the file is not there or has no such line, or the evidence is a never-opened file or a path outsideroot(step 2).
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.
- 2d ago First seen · 94 lines · 70 tokens per session scan A 1d78e7d0d5c5
system-verifier is an agent published in the GitHub repository ApurvBazari/claude-plugins (1 stars, last pushed 2d ago), licensed MIT. It adds 70 tokens to every session and 1,383 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 2026-10-14.
Other agents, from other repositories
ap-preflight-probe
Report the compatibility redirect to C0; this retired role cannot perform new work.
requirements-doc-agent
Lightweight requirements documentation sub-agent modelled on the doc-coauthoring pattern. Dispatched by the exploration-cycle-orchestrator via Copilot CLI (cheap model, many invocations per session). Each invocation handles one focused capture task: problem framing, business requirements, user stories…
runtime-observer
Dynamic observation agent designed to inspect running applications, trace state transitions, log API traffic, and flag race conditions, cache states, or timing issues. Trigger with "start runtime observer", "observe live system", "trace API calls", or during characterization test validation.
debate-synthesizer
Multi-Agent Debate Judge. Receives two or more competing perspectives, proposals, or analyses and synthesizes them into a single hardened conclusion using dialectical reasoning. Resolves conflicts, names tradeoffs, and produces a final verdict.
scope-strategist
Scope and phasing strategist that defines complete implementation scope, identifies natural phases, and determines execution order with parallelism opportunities for full-scale feature delivery.
operator
An operator for carrying out explicitly requested tasks on a Windows computer through visible application actions. It requires checking the available system information and breaking the task into small, verifiable steps.