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/rohitgehe05/mindpowers/fact-checkingnpx skills add rohitgehe05/mindpowers --skill fact-checkinggit clone --depth 1 https://github.com/rohitgehe05/mindpowersWrote 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/rohitgehe05/mindpowers/fact-checking)<a href="https://agentmods.dev/skills/rohitgehe05/mindpowers/fact-checking"><img src="https://agentmods.dev/badge/skills/rohitgehe05/mindpowers/fact-checking.svg" alt="Measured on agentmods" 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 | $0.00113 | $0.01590 |
| Opus 5 | $0.00056 | $0.00795 |
| Sonnet 5 | $0.00023 | $0.00318 |
| Haiku 4.5 | $0.00011 | $0.00159 |
Grade A, and why
fact-checking 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 4d 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 — 143 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Fact-Checking
Overview
The last pass before a doc ships. Two jobs: (1) verify every checkable claim against the sources actually in scope, with verdicts that must quote their source; (2) flag content that could damage someone if shipped — named people, customer names, unannounced plans, confidential bleed, PII.
This is the "fact-check" step of the mindpowers loop (validate → shape →
draft → review → fact-check → remember). It runs on demand and is
recommended — never forced — by drafting and reviewing-docs at handoff.
The gate is honest reporting, not blocking: the doc always comes back to the
user, with the unresolved list on top. Track the steps below as todos if
your harness has a task list.
Step 1: Extract claims
Read the doc and extract every checkable claim: numbers, dates, named
people and organizations, quotes, "data shows X" statements, superlatives
("first", "only", "biggest"). Each claim gets a stable ID at first
sighting: C-001, C-002, …, scoped to the doc and never renumbered.
Before assigning IDs, check docs/mindpowers/reviews/ for a prior review
or fact-check of this doc (same filename stem). If a claims ledger exists,
existing claims keep their C-###, new claims continue the sequence, and
this re-check updates verdicts under existing IDs.
Step 2: Classify
Classify each claim:
- source-verifiable — a source inside scope could settle it. In scope: files or links the user supplied or pointed at, the locked spec, the problem brief. Nothing else.
- external — it lives in someone's head, a dashboard, or a record not in scope. Structurally uncheckable here.
- unverifiable — no source could settle it as stated (vague superlatives, unfalsifiable claims).
Honesty rule: when in doubt between source-verifiable and external, classify external. "Confirm this with the source" beats "looks right."
If a decision-critical claim is public (a market size, a public date, a public quote), offer web research as an option and wait for the user's yes. Never run it automatically.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 4d ago First seen · 143 lines · 113 tokens per session scan A e63b55af836a
fact-checking is a skill published in the GitHub repository rohitgehe05/mindpowers (5 stars, last pushed 27d ago), licensed MIT. It adds 113 tokens to every session and 1,590 once invoked, about $0.0006 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
anti-ai-writing
Use when writing or editing any prose for a human reader - documentation, READMEs, emails, reports, captions, scripts, blog posts, UI copy, commit messages, PR descriptions - or when asked to remove AI tells, de-slop text, make writing sound human, or check writing style. Applies 29 researched patterns that make text…
markifact-overview
Reference — what Markifact is, what the MCP server exposes, and the discover→inspect→run pattern. Always loaded into the performance-marketer agent.
safe-write-operations
Reference — rules for safely executing write/destructive operations against ad accounts. Always loaded into the performance-marketer agent.
anti-slop-writing
Make writing sound like a person wrote it. Rewrites drafts that read as machine-generated, restores voice that AI editing flattened, and audits text for AI tells when asked. Use when a draft feels generic or corporate, when asked to de-slop or humanize text, when polishing anything an LLM helped write, before…
technical-content-writer
Write technical content that sounds like a specific human wrote it - LinkedIn posts, blog posts, threads, newsletters, essays about engineering and infrastructure subjects. Derives a voice signature from exemplars before drafting, forces mechanism and numbers over adjectives, and runs an anti-slop gate before output.…
dev
Software development planning. Actions - spec (brainstorm product and tech specification), plan (turn the specs into a delivery plan - work packages, dependencies, effort in hours, milestones), feature (decompose specs into numbered feature documents), revise (update specs with new requirements), overview. Planning…