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 skills add twaldin/flt --skill grillgit clone --depth 1 https://github.com/twaldin/fltWrote 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/skills/twaldin/flt/grill)<a href="https://agentmods.dev/skills/twaldin/flt/grill"><img src="https://agentmods.dev/badge/skills/twaldin/flt/grill/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/skills/twaldin/flt/grill"><img src="https://agentmods.dev/badge/skills/twaldin/flt/grill.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 | $0.00066 | $0.01549 |
| Opus 5.5 | $0.00026 | $0.00620 |
| Sonnet 5.5 | $0.00013 | $0.00310 |
| Haiku 4.5 | $0.00007 | $0.00155 |
Grade A, and why
grill 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 — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Grill
You are a senior engineer doing a design review. The user has a feature, bug fix, or code change they want to make. Your job is to ask every question that needs answering before implementation can begin — edge cases, failure modes, behavior under unusual conditions, scope boundaries, naming, data flow, state management, interactions with existing code.
Why this exists
The user knows from experience that jumping straight to code leads to rework. They want you to be the person who says "wait, what happens when..." before a single line is written. Your questions should surface the decisions that would otherwise be discovered mid-implementation and cause backtracking.
Process
1. Research first, then question
Before asking anything, silently research the codebase to understand the context around the user's request. Read relevant files, trace data flows, understand existing patterns. The quality of your questions depends entirely on how well you understand what's already there.
Use Explore agents or direct file reads — whatever gets you the context you need. Don't tell the user you're researching; just do it, then come out swinging with informed questions.
2. Enter plan mode
After your research, enter plan mode. This is where the entire Q&A happens. You'll build up the implementation plan as answers come in.
3. Ask questions via the structured-question primitive — never wall-of-text
Dispatch rule (check at runtime):
- If
fltis on PATH AND~/.flt/exists, useflt ask human '<json>'via the Bash tool. This persists the Q/A under~/.flt/qna/so the mutator/GEPA loop can train on it. The JSON shape is identical to AskUserQuestion's (1-4 questions, 2-4 options each,multiSelect, optionalpreviewfor single-select). Block on the bash call until the human answers; the command prints the answer JSON to stdout. - Otherwise fall back to the native
AskUserQuestiontool. Same JSON shape, same batching rules.
Either way, never dump prose with "Q1/Q2/Q3" sections — that's less readable than the interactive picker.
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 First seen · 107 lines · 66 tokens per session scan A f08578c51fc1
grill is a skill published in the GitHub repository twaldin/flt (5 stars, last pushed 1mo ago), licensed MIT. It adds 66 tokens to every session and 1,549 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 2026-08-31.
Other skills, from other repositories
harness-init-runner
Initialize a lightweight repo-local Node.js harness (harness/ + .harness/) WITHOUT AIOS dependency. Use ONLY when you need a standalone, portable harness. If AIOS is installed, use aios-long-running-harness instead — it has rex Command hosting, ContextDB integration, and checkpoint recovery.
searxng-websearch
Use this skill whenever the user wants to search the web, do research on a topic, fetch a webpage, or gather information from online sources using a self-hosted SearXNG instance. Triggers include: 'search for', 'look up', 'find online', 'deep research', 'fetch this page', 'web search', or any request that needs…
Interceptor
Real Chrome/Brave + macOS Computer Use from inside the browser — zero CDP fingerprint, real sessions; mandatory for visual deploy verification. Drive clickwork yourself. USE WHEN verify deploy, confirm UI, screenshot verification, computer use, macos automation, debug web, troubleshoot, visual check, motion/animation…
Art
Static visual content across 20+ formats — diagrams, mermaid, infographics, D3 dashboards, comics, icons, wallpaper — via Flux, Nano Banana Pro, and GPT-Image-2. USE WHEN art, illustration, diagram, flowchart, infographic, header image, blog social thumbnail, visualize, generate image, mermaid, architecture diagram…
ISA
Owns the Ideal State Artifact — the primitive holding a project or task's articulated ideal state; scaffolds, interviews, scores completeness, reconciles feature excerpts to master, seeds from a repo, and appends decisions/changelog/verification across a locked fourteen-section order. USE WHEN ISA, ISC, ideal state…
Research
Multi-agent web research with mandatory URL verification, confidence-tagged output, and four depth modes (quick to deep investigation). USE WHEN research, do research, quick research, extensive research, deep investigation, find information, investigate, extract alpha, analyze content, retrieve content, AI trends…