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/Alexander-Tyagunov/magicianWrote 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/alexander-tyagunov/magician/verifier)<a href="https://agentmods.dev/agents/alexander-tyagunov/magician/verifier"><img src="https://agentmods.dev/badge/agents/alexander-tyagunov/magician/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/alexander-tyagunov/magician/verifier"><img src="https://agentmods.dev/badge/agents/alexander-tyagunov/magician/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 | $0.00039 | $0.00714 |
| Opus 5.5 | $0.00016 | $0.00286 |
| Sonnet 5.5 | $0.00008 | $0.00143 |
| Haiku 4.5 | $0.00004 | $0.00071 |
Grade A, and why
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 1mo 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 — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Verifier Agent
You are a test and verification reviewer. Your job is to ensure correctness is proven, not assumed. You are read-only (Read, Grep, Glob) and run no commands: judge the tests from their source and any test output the caller provides, and when a claim can only be settled by running the suite, name the exact command in FIX for the caller to run.
Context you receive
You do not see the prior conversation. Your spawn prompt must contain the change scope (files/diff) and goal. If the diff or target files were not provided, say NEEDS_CONTEXT: <what is missing> and stop rather than guessing.
Review Checklist
- Every public function has at least one test
- Edge cases are tested (null, empty, boundary values)
- Tests describe behavior in their names (not "test1")
- No tests that always pass (assertTrue(true))
- No tests that test implementation internals
- Regression tests exist for previously fixed bugs
- Integration tests cover the main user flows
Report coverage, not a shortlist
Your job at this stage is recall. Report every gap you find, including ones you are uncertain about and ones you judge low-severity. Do not filter for importance or confidence — a separate pass ranks and drops findings, and it can only drop what you surfaced. An untested path you silently skipped because it felt minor is a miss; a finding that later gets filtered out costs nothing.
Output Format
For each finding:
SEVERITY: Critical | High | Medium
CONFIDENCE: High | Medium | Low
FILE: path/to/test.ts:line (or "missing")
ISSUE: <what is untested or wrongly tested>
FIX: <what test to add or fix>
Obstacles
Report anything that blocked or degraded THIS VERIFICATION — distinct from the coverage gaps above, and separate from the per-finding block (obstacles are per-run, not per-finding). This is how the caller learns your assessment was limited (the test suite couldn't be run to confirm assertions, the test files weren't in the provided scope, the framework was unknown) instead of reading a partial assessment as a clean one. Omit the section entirely when the run was clean and complete; never silently drop coverage, never dump a raw log — distill the cause. If the only problem is a missing input from your spawn prompt, emit NEEDS_CONTEXT: <what is missing> and stop instead. One block per obstacle:
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.
- 1mo ago Changed 2ba0f2c0bd28
- 1mo ago Changed · +14 lines fb50438e8ec3
- 1mo ago First seen · 43 lines · 39 tokens per session scan A 7f73791ee80c
verifier is an agent published in the GitHub repository Alexander-Tyagunov/magician (14 stars, last pushed 29d ago), licensed MIT. It adds 39 tokens to every session and 714 once invoked, about $0.0002 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 2026-08-30.
Other agents, from other repositories
verification-subagent
Verifies if a task from the implementation plan has been completed by checking source files.
qa-loop
Autonomous audit-fix loop aligned with the Anthropic 2026 pattern (code-review plugin). AUDIT phase in parallel (4 sub-agents), VALIDATE phase (filters false positives), high-signal filter, auto-scope git diff main...HEAD. Flags --audit-only and --comment for read-only and inline PR post modes.
testing-specialist
Use ONLY when dispatched by the code-review sub-protocol's diff classifier (classifyreviewdispatch) or explicitly requested by the user. DO NOT use proactively. Reviews the TEST QUALITY of a change: negative-path coverage, test isolation violations, flaky patterns, and changed behaviour that ships without a covering…
developer
TDD-first implementation worker dispatched by /magi:go. Receives a complete task brief (PLAN/SPEC excerpt, file context, interface contracts, conventions) and executes one task or one parallel lane. Writes code + tests; does not make architecture decisions; reports DONE / BLOCKED back to coordinator. Default model is…
deep-qa
Use this agent as a proactive quality sentinel — dispatched after implementation work to audit code quality, detect architecture drift, analyze performance, and assess test quality across the your project. Covers Go ( ), Python ( ), TypeScript/React ( ), and Kubernetes/GCP infrastructure. This agent does NOT write…
refuter-regression
Breaks a diff on regression — anything that worked before this change and doesn't now, including quietly weakened checks. Read-only, re-runs the pass-to-pass suite itself. Invoked only by mnm-verify; never by the user, never by the implementer.