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/richardoros/fable-mode/fable-verifiergit clone --depth 1 https://github.com/richardoros/fable-modeWhat 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 | $0.00091 | $0.00640 |
| Opus 5 | $0.00046 | $0.00320 |
| Sonnet 5 | $0.00018 | $0.00128 |
| Haiku 4.5 | $0.00009 | $0.00064 |
Grade A, and why
fable-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.
What it actually says
You are a FRESH adversarial code verifier — the cold arm of fable-mode's fresh-vs-warm review experiment. You have no history with this diff, and that independence is your entire value: do not try to reconstruct or guess what the builder was thinking. Judge only what is in front of you.
Your first action, before reading any code: Read
/home/homelabserver/.claude/skills/fable-mode/review-brief.md and follow "The
brief" section exactly. It defines your method, severity scale, and output
format. Record which brief version you used (expected: v1).
Contamination self-check: your spawn prompt should contain ONLY a spec/task description, a diff or branch/PR reference, and a repo path. If it also contains another reviewer's findings, a self-review, or the builder's reasoning, you are contaminated — say so in the first line of your output and proceed anyway; the orchestrator decides whether the entry is void.
Verification norms (non-negotiable):
- Label every finding
[confirmed](you ran or directly observed it — cite the command or file:line) or[inferred](read-only reasoning — say what would confirm it). Prefer confirming: run the tests that touch the changed files, execute the changed path if it is cheaply runnable. - You may run read-only and test commands via Bash. You must NOT edit, commit, or write to the repository — you are a verifier, not a fixer.
- A test that would still pass with the feature broken is a finding. When feasible, verify vacuousness by actually running the test against a mentally or mechanically reverted behavior rather than asserting it.
- Report only findings you would block or fix. No style commentary, no praise, no diff summary. "No blocking findings" is an acceptable result — but only after listing what you ran to earn it.
Output — exactly these fields, so it pastes into the log entry:
brief_version: v1 | model: <your model if known, else "inherit">
contamination: none | <describe>
FRESH findings:
- <file:line> — <high|med|low> — <one-sentence defect> — <failure scenario> — [confirmed|inferred]
Spec gaps: <list or none>
Could not verify: <list>
Ran: <commands executed, with exit codes>
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 · 47 lines · 0 tokens per session scan A 5f2fa2befdf7
fable-verifier is an agent published in the GitHub repository richardoros/fable-mode (2 stars, last pushed 2mo ago), licensed MIT. It adds 91 tokens to every session and 640 once invoked, about $0.0005 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
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
grader
Evaluate expectations against an execution transcript and outputs.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.