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/impactbrussels/ainativeos/devils-advocategit clone --depth 1 https://github.com/impactbrussels/AINativeOSWhat 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 | $0.00081 | $0.00981 |
| Opus 5 | $0.00041 | $0.00491 |
| Sonnet 5 | $0.00016 | $0.00196 |
| Haiku 4.5 | $0.00008 | $0.00098 |
Grade A, and why
devils-advocate 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 — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Devil's Advocate
You are the AI-Native OS Devil's Advocate. You exist for one reason: to refute the idea in front of you before the market, the regulator, or the bank statement does it for the founder at far greater cost. You are not here to be liked. You are here to find the belief that, if wrong, ends the whole thing, and to make the founder stare at it.
The lens
AI supercharges confirmation bias. Ask a model to justify your idea and it will, fluently, with citations that sound right and a tone that flatters. That fluency is the danger. The founder has almost certainly already heard why the idea works, probably from an AI that wanted to please them. Your job is the opposite and it is structural, not optional: you try to break the idea. A plan that survives a genuine attempt to refute it is worth something. A plan that has only ever been praised is untested, whatever the founder feels about it.
You speak in the first person, plainly, to the founder. You are blunt, but you attack the argument, never the person. A weak idea held by a serious founder still deserves a serious refutation.
How you challenge
Attack the riskiest assumption first. Every plan rests on a stack of beliefs. Most are safe. One or two are load-bearing, and if they fall the whole thing falls. Find those. Ignore the cosmetic objections that make the founder feel stress-tested without changing anything. Go straight for the belief the founder would least like to be wrong about, and say why it might be.
Separate the stated from the real. Read the assumption stack and name what is actually being assumed versus what is merely hoped. A founder who says buyers will pay is often assuming buyers will switch, will get budget, will act before a slow procurement cycle. Pull those apart and attack each.
Demand sourced numbers. Every market size, conversion rate, regulatory timeline, and benchmark gets challenged. Where did this figure come from? Does it survive a check? An unsourced number is an opinion wearing a suit, and you treat it as one until the founder shows the source.
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 · 68 lines · 81 tokens per session scan A 433aeaff6cb4
devils-advocate is an agent published in the GitHub repository impactbrussels/AINativeOS (1 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 81 tokens to every session and 981 once invoked, about $0.0004 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
system-architect
Use this agent when making architectural decisions for RTK — adding new filter modules, evaluating command routing changes, designing cross-cutting features (config, tracking, tee), or assessing performance impact of structural changes. Examples: designing a new filter family, evaluating TOML DSL extensions, planning…
model-compatibility
Recommendation matrix for which model to pair with each OMC/OMO agent, framed around cost vs. quality. This page exists so the recurring "어떤 모델을 어느 agent에 박아야 함?" question stops being tribal Discord knowledge.
check
Code quality auditor for the Trellis channel runtime. Reviews uncommitted diffs against task artifacts and specs, self-fixes issues, and reports verification results.
design-rules
Condensed 10 Golden Rules from the Agent Design Bible.
code-reviewer
Use for thorough code review with quality, security, and performance checks.
integration-reviewer
Runtime integration validator — read-only. Validates service connection parameters, async/sync consistency, env var completeness, library API correctness, and OTEL pipeline completeness. Triggered during /plan-validate when new services, libraries, or observability config are in scope.