A Reality Check workflow for analysing an article or other source, extracting its claims, storing them in a database, and validating the analysis. bioRxiv, mentioned by some source-analysis tasks, is a website where researchers share scientific papers before formal peer review.

Use it to fetch a source, analyse it in stages, register claims in the Reality Check data repository, and run validation.

It provides a defined process for checking source claims instead of leaving the analysis as unstructured notes.

check — Skill for Claude Code and Codex by lhl/realitycheck

Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo

git clone --depth 1 https://github.com/lhl/realitycheck

agentmods

npx agentmods add skills/lhl/realitycheck/check

Made for: Claude Code, Codex.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/lhl/realitycheck/check/github.svg)](https://agentmods.dev/skills/lhl/realitycheck/check)

Your own site

<a href="https://agentmods.dev/skills/lhl/realitycheck/check"><img src="https://agentmods.dev/badge/skills/lhl/realitycheck/check/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/lhl/realitycheck/check"><img src="https://agentmods.dev/badge/skills/lhl/realitycheck/check.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 34 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 9,002 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

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	$0.00034	$0.09002
Opus 5.5	$0.00014	$0.03601
Sonnet 5.5	$0.00007	$0.01800
Haiku 4.5	$0.00003	$0.00900

Measured 1mo ago against content hash 5c55e2f61e5a, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

check scanned grade A with 1 finding 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- Alternative: `curl -L -sS "URL" | rc-html-extract - --format json`

integrations/amp/skills/check/SKILL.md · 988 lines

How it starts

The opening of the file, as written. The whole thing — 988 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Full Analysis Workflow

Full Reality Check analysis - fetch source, perform 3-stage analysis, extract claims, register to database, and validate. The flagship command for rigorous source analysis.

When This Skill Activates

"Analyze this article for claims"
"Reality check this URL"
"Run a check on"
"Perform a full analysis"

The flagship Reality Check command for rigorous source analysis.

Prerequisites

Environment

Set REALITYCHECK_DATA to point to your data repository:

export REALITYCHECK_DATA=/path/to/realitycheck-data/data/realitycheck.lance

The PROJECT_ROOT is derived from this path - all analysis files go there.

CLI Commands

Reality Check provides CLI tools (rc-db, rc-validate, rc-export, rc-embed).

Check availability:

which rc-db  # Should show path if pip-installed

If commands are not found, either:

Install: pip install realitycheck (recommended)
Use uv run from framework directory: uv run python scripts/db.py ...
Add framework as submodule and use: .framework/scripts/db.py ...

Red Flags: Wrong Repository

IMPORTANT: Always write to the DATA repository, never to the framework repository.

If you see these directories, you're in the framework repo (wrong place for data):

scripts/
tests/
integrations/
methodology/

Stop and verify REALITYCHECK_DATA is set correctly.

Data Source of Truth

LanceDB is the source of truth, not YAML files.

Query sources: rc-db source get <id> or rc-db source list
Query claims: rc-db claim get <id> or rc-db claim list
Search: rc-db search "query"

Ignore YAML files like claims/registry.yaml or reference/sources.yaml - these are exports/legacy format.

Workflow Steps

Start Tracking - Begin token usage capture (lifecycle mode)
Fetch - Retrieve and parse source content
- Primary: WebFetch for most URLs
- Alternative: curl -L -sS "URL" | rc-html-extract - --format json
- rc-html-extract returns structured {title, published, text, headings, word_count}
- Use the extract tool when you need clean metadata or main text extraction
Metadata - Extract title, author, date, type, generate source-id
Stage 1: Descriptive - Neutral summary, key claims, argument structure
Stage 2: Evaluative - Evidence quality, fact-checking, disconfirming evidence
- DB-first for crux factuals: run rc-db search "<neutral keywords>" before external search
- Web discovery: use your environment's web search capability (e.g., Claude WebSearch, Codex web.run search_query)
- Minimum verification attempt: for each crux [F] claim, run >=2 distinct queries and record attempts
- Status contract: use ok, x, nf, blocked, ? in "Key Factual Claims Verified"
- Review gate: do not mark [REVIEWED] if any crux [F] claim remains ?
Stage 3: Dialectical - Steelman, counterarguments, synthesis
Neutral Analyst Prose Pass - Clean up analyst-authored prose without changing source content, quotes, claim IDs, evidence levels, or credences
Extract - Format claims as YAML
Register - Add source and claims to database
Provenance (for high-credence claims) - Link evidence + capture reasoning trails
Complete Tracking - Finalize token usage + register analysis_logs row
Validate - Run integrity checks
README - Update data project analysis index
File Inbox - Move/archive inbox items to permanent locations
Commit - Stage and commit changes to data repo
Push - Push to remote
Report - Generate summary

Read the full file on GitHub · 988 lines

Changes

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 · 988 lines · 34 tokens per session scan A 5c55e2f61e5a

Subscribe to this mod's changes

check is a skill published in the GitHub repository lhl/realitycheck (44 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 34 tokens to every session and 9,002 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Analytics and metrics Research agent analysis

Deep Research

Produce a deep, structured research report on a topic: decompose into key dimensions, analyze each with evidence and reasoning, synthesize cross-cutting insights, and surface open questions. Use for deep research, analysis, and literature/landscape reviews.

AgentEra/Agently · 53 tokens

check