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 thoughtbot/rails-consultant --skill hard-newsgit clone --depth 1 https://github.com/thoughtbot/rails-consultantWrote 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/thoughtbot/rails-consultant/hard-news)<a href="https://agentmods.dev/skills/thoughtbot/rails-consultant/hard-news"><img src="https://agentmods.dev/badge/skills/thoughtbot/rails-consultant/hard-news/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/thoughtbot/rails-consultant/hard-news"><img src="https://agentmods.dev/badge/skills/thoughtbot/rails-consultant/hard-news.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 | $0.00022 | $0.00631 |
| Opus 5 | $0.00011 | $0.00316 |
| Sonnet 5 | $0.00004 | $0.00126 |
| Haiku 4.5 | $0.00002 | $0.00063 |
Grade A, and why
hard-news 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 12d 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 — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Behavior
This is a prep session. The goal is to walk into a hard conversation clear-headed, not rehearsed. Start by understanding the situation conversationally — one question at a time.
Open with:
"Tell me what happened — the version you'd say to a colleague, not the client. Don't soften it yet."
Wait for their answer. Then ask what matters most for shaping the conversation:
"What do you know about how this client handles bad news? And what do you need from this conversation — what's the outcome you're steering toward?"
Wait for their answer. Then, before preparing anything:
"Write your opening two sentences right now — the actual words you'd say to start this conversation. Don't polish them, don't second-guess them. Just write what comes out."
Wait for their draft. This is the most important step — it reveals how they instinctively handle pressure. Open the prep with honest feedback on their draft: what's working, what would land badly, and what to change.
Then deliver the full prep. Keep it tight — they need to be able to hold this in their head going into the conversation:
The news in one sentence — the clearest, most honest version. No jargon, no hedging, no burying the lead.
How to open — the first 3–4 sentences verbatim. Lead with the news, not a preamble. Bad news after small talk feels like manipulation. Name the problem directly, take appropriate accountability without over-apologising, signal there's a plan.
The explanation — how to describe what happened in plain terms the client understands. What happened, why, what you know vs. what you're still figuring out, and what it means for the project.
The path forward — what you're proposing. Specific, not vague. If the plan isn't complete yet, say what you do know and when you'll have more.
Likely reactions — 2–3 reactions given what you know about this client, and how to respond to each. Non-defensive, grounded. Defensiveness signals you care more about being right than solving the problem.
What not to say — 2–3 specific phrases or moves to avoid. Things that feel natural under pressure but damage trust: over-apologising, deflecting blame, minimising impact, over-promising.
Close with:
"What did your unpolished opening reveal about how you instinctively handle discomfort? Did you soften, deflect, or lead straight?"
Wait for their answer. Respond with one short paragraph: what their draft pattern reveals about their default communication instinct under pressure, and the one habit worth practising.
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.
- 12d ago First seen · 49 lines · 22 tokens per session scan A a37acccce009
hard-news is a skill published in the GitHub repository thoughtbot/rails-consultant (24 stars, last pushed 1mo ago), licensed MIT. It adds 22 tokens to every session and 631 once invoked, about $0.0001 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 2026-08-30.
Other skills, from other repositories
recursive-decomposition
Decompose dense codebase-wide, multi-document, PDF, and aggregation work even when the input fits the context window, following Recursive Language Models (Zhang, Kraska, Khattab, 2025). Use when the user asks to analyse all files, a whole repo, all docs, large PDFs, or to aggregate or multi-hop across scattered…
swarm
Run a multi-agent audit of a codebase by spawning specialized parallel subagents (security, performance, tests, architecture, dead-code), then synthesize their findings into a single prioritized action plan. Use this whenever the user runs /swarm, asks to "audit the repo," "review this codebase," "find issues across…
full-security-review
Structured security audit covering injection, auth, secrets, input validation, dependencies, cryptography, and AI/LLM risks. Produces severity-graded findings.
log-correlation
Correlates and troubleshoots logs across OS (Linux/macOS), AWS (CloudWatch, CloudTrail, ALB, Lambda), application (JSON, logfmt), and web servers (Nginx, Apache).
tools-unity-behavior-designer
Behavior Designer patterns for AI behavior trees including task creation, shared variables, conditionals, and debugging.
tools-unity-flowcanvas
FlowCanvas visual scripting patterns for abilities, custom nodes, and graph execution.