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 skills/everywan-dev/claude-code-engineering/get-a-second-model-opinionnpx skills add everywan-dev/claude-code-engineering --skill get-a-second-model-opiniongit clone --depth 1 https://github.com/everywan-dev/claude-code-engineeringWhat 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 | $0.00042 | $0.01809 |
| Opus 5 | $0.00021 | $0.00905 |
| Sonnet 5 | $0.00008 | $0.00362 |
| Haiku 4.5 | $0.00004 | $0.00181 |
Grade A, and why
get-a-second-model-opinion 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 — 182 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Get a second model opinion
Asking yourself again is not a second opinion. Whatever made the first answer wrong is still in place for the second one: the same priors, the same reading of the same evidence, the same thing you failed to consider. People have this problem too, which is why review is done by somebody else.
So when a claim needs checking and no command can settle it, take it to a different model — a different family, a different provider, something with a different training history. Not the same one in a fresh session.
When to use this
- A conclusion is load-bearing and nothing you can execute will confirm it
- You are about to write something into long-lived documentation
- A design decision hinges on a fact you are confident about but never verified
- You have been reasoning about the same problem long enough to be attached to your explanation
- Two of your own hypotheses fit the evidence equally well
What this can and cannot answer
The second model has no access to your systems. It cannot see your files, run your commands, reach your hosts or read your logs. That single fact decides the whole question taxonomy.
| Ask it | Do not ask it |
|---|---|
| Is this general claim true? | Is this service running? |
| What is wrong with this reasoning? | Does this file contain X? |
| What does this design assume that nobody stated? | Did the change apply? |
| What is a second explanation for these symptoms? | Which version is deployed? |
| Which of these two readings of this specification is right? | Is this host reachable? |
| What is the failure mode nobody listed here? | Why did this command fail on my machine? |
🔴 Anything whose answer lives inside your environment is not a cross-check question. Sent anyway, you will get a confident, fluent, entirely invented answer — because a plausible answer is the only thing that can be produced when the real one is unreachable.
The rule of thumb: if the answer could be obtained by running something, run it. A cross-check is for the questions no command settles.
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 · 182 lines · 42 tokens per session scan A cd1e9105b8bf
get-a-second-model-opinion is a skill published in the GitHub repository everywan-dev/claude-code-engineering (2 stars, last pushed 14d ago), licensed Apache-2.0. It adds 42 tokens to every session and 1,809 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 skills, from other repositories
claude-project-manifest
Creates annotated bibliography-style manifests for Claude projects, tracking files, conversation threads, and relationships with unique IDs and annotations. Use when documenting project contents, creating file inventories, tracking conversation history, or building navigable knowledge maps.
proofpress
Preserve or inspect decision history for Markdown and static HTML knowledge artifacts with Proofpress. Use when creating or accepting a meaningful revision to an ADR, design doc, PRD, README, strategy note, portable .md artifact, or static .html/.htm artifact in a configured repository, or when receiving a file…
llm-wiki
Maintain a personal team knowledge base using the LLM Wiki pattern — incremental ingest, query, and lint operations on a layered wiki architecture.
PBE Extractor
Extract invariant principles from any text — find the ideas that survive rephrasing.
Principle Synthesizer
Synthesize invariant principles from 3+ sources — find the core that survives across all expressions.
wiki-lint
Run a deterministic, read-only health check on an Obsidian wiki. Use for lint, vault health check, audit wiki health, find orphans, find dead links, frontmatter audit, provenance audit, or wiki audit. Reports graph, link, frontmatter, provenance-ledger, empty-section, and stale-index findings; it does not reason…