New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Factcheck Guard

by vibeeval in vibeeval/vibecosystem

A skill your agent uses when making any factual claim about the codebase — existence, absence, or behavior.

MITAuto-check passedResearch & Science

Install Factcheck Guard

skills CLI
$ npx skills add vibeeval/vibecosystem --skill factcheck-guard -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install vibeeval/vibecosystem factcheck-guard --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/vibeeval/vibecosystem.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/factcheck-guard .claude/skills/factcheck-guard && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
factcheck-guard
GitHub stars
532
Token cost
~2.2k tokens
SKILL.md length
652 words
Files
1
Skills in repo
12
Repo updated
First seen
Licence
MIT

At a glance

A skill your agent uses when making any factual claim about the codebase — existence, absence, or behavior.

  • Making any factual claim about the codebase — existence
  • SKILL.md covers Why This Exists, Confidence Markers (MANDATORY), Verification Protocol and Common False Claim Patterns, plus 4 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
  • Tasks that involve Fact-checking and source verification

What it does

Factcheck Guard is an agent skill from vibeeval/vibecosystem. Use this skill when making any factual claim about the codebase — existence, absence, or behavior. Converts the claim-verification rule into a systematic action protocol that prevents false assertions from grep-only results.

Its SKILL.md is about 2.2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Research & Science, covering Fact-checking and source verification. The repository describes itself as: AI software team for Claude Code - 138 agents, 295 skills, 73 hooks. Self-learning, multi-agent swarm, autonomous skill evolution. The licence is MIT.

When your agent uses it

  • Making any factual claim about the codebase — existence
  • Tasks that involve Fact-checking and source verification

Example prompts

  • “/factcheck-guard”

What it can do on your machine

Read from SKILL.md and the folder at commit 3b763b1. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Factcheck Guard loads about 2.2k tokens when it runs. Until then it costs about 60 tokens; SKILL.md has 652 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~60
When it runs · the whole SKILL.md, loaded when a task matches
~2.2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from vibeeval/vibecosystem at commit 3b763b1, republished under its MIT licence (© vibeeval). 652 words, ~2,177 tokens.

Download SKILL.mdSave it as .claude/skills/factcheck-guard/SKILL.md (or your agent's skills folder).
name
factcheck-guard
description
Use this skill when making any factual claim about the codebase — existence, absence, or behavior. Converts the claim-verification rule into a systematic action protocol that prevents false assertions from grep-only results.

FactCheck Guard — Runtime Claim Verification

Prevents false claims about the codebase. Grep results are hypotheses, not proof. Every assertion about "what the code does" must be verified by reading the actual code.

Why This Exists

An 80% false claim rate was observed when agents trusted grep results without reading files. The pattern repeats: grep finds a string, agent asserts behavior, behavior is wrong. This skill converts the claim-verification guidelines into an actionable protocol.

Confidence Markers (MANDATORY)

Every factual claim about the codebase MUST carry one of these markers:

MarkerMeaningRequirements
[VERIFIED]Read the file, traced the logicMust cite file:line
[INFERRED]Based on grep or file nameMust verify before asserting to others
[UNCERTAIN]Not yet checkedMust investigate before making the claim

Never promote an [INFERRED] claim to a statement of fact. Never let [UNCERTAIN] appear in a response to the user — investigate first.

Verification Protocol

Level 1: Existence Claims

When asserting "X exists in the codebase":

Step 1 — Search (produces [INFERRED])

Grep/Glob for the pattern.
Result: "X appears at path/to/file.ts:42" [INFERRED]

Step 2 — Read the file (required to reach [VERIFIED])

Read path/to/file.ts, locate line 42.
Confirm the match is actual code, not a comment or string literal.

Step 3 — Check context

Is the code in a live code path, or inside a dead branch / disabled block?
Is it imported and actually used?
Does the function name match what you expected, or is it a different concept?

Step 4 — State the claim with citation

"Error handling exists at src/api/handler.ts:67 — try/catch wraps the database call" [VERIFIED]
Level 2: Absence Claims

When asserting "X does NOT exist" — absence claims are the most dangerous. A single missed search term invalidates the entire claim.

Step 1 — Primary search

Grep for the primary term. No results found.
Status: [INFERRED ABSENT] — not yet VERIFIED

Step 2 — Alternative terms (minimum 3)

Different naming conventions, abbreviations, synonyms.

Example: Searching for "rate limiting"
Try: "rateLimit", "rate_limit", "throttle", "throttling", "requestLimit", "limiter"
If any match: investigate that file before concluding

Step 3 — Check likely locations manually

Where would X logically live if it existed?
Read those files directly — grep may miss dynamically constructed patterns.

Example: Checking for auth middleware
Read: src/middleware/, app/middleware.ts, server.ts, app.ts
Even if grep found nothing, scan these files for auth-related logic.

Step 4 — Only then claim VERIFIED ABSENT

"Rate limiting does not exist. Searched: rateLimit, rate_limit, throttle, throttling,
requestLimit, limiter. Checked: src/middleware/index.ts, server.ts, app.ts. None found." [VERIFIED ABSENT]
Level 3: Behavior Claims

When asserting "the system does X" — the most subtle failure mode. Finding a function name does not tell you what the function actually does.

Step 1 — Locate the function

Grep/Glob finds the function name.
Status: [INFERRED] — you know it exists, not what it does

Step 2 — Read the full function body

Read the file. Read the entire function, not just the signature.
Look for: branches, early returns, side effects, what it actually returns.

Step 3 — Trace the logic path

Follow the execution path relevant to your claim.
If the function calls other functions, read those too (one level deep minimum).
Check: what happens in the error path? What are the edge cases?

Step 4 — State behavior with evidence

"processPayment() validates card number via Luhn check (payment.ts:34),
then calls Stripe API (payment.ts:41), and on failure rolls back the DB
transaction (payment.ts:58) — it does NOT charge the card on validation failure." [VERIFIED]

Common False Claim Patterns

PatternWhy It FailsCorrect Approach
grep found "try.*catch"Could be in comment, string, or test mockRead the file, confirm it wraps real code
grep found nothingWrong term, wrong directory, different namingTry 3+ alternative terms
File named auth.ts existsDoesn't mean auth is implementedRead the file
Type definition has a fieldDoesn't mean the field is set or usedGrep for field assignment and usage
Function is called somewhereDoesn't mean it does what the name impliesRead the function body
Found in tests directoryTest mocks are not real implementationsConfirm the actual source file
grep -L returned filesThose files lack the pattern — but pattern might exist under a different nameTry alternative terms before concluding
Found in commented-out blockNot active codeCheck if block is enabled
Show full SKILL.md (230 more words)Show less

Two-Pass Audit Protocol

For system audits, security reviews, and architecture reviews:

Pass 1: Generate Hypotheses

Go through each area systematically. Produce hypothesis statements only:

"Input validation might be missing from the /api/upload route" [INFERRED]
"The auth token might not be refreshed on expiry" [INFERRED]
"Database queries might not use parameterized statements" [INFERRED]

Do not state these as facts. Do not include them in the user-facing summary.

Pass 2: Verify Each Hypothesis

For every [INFERRED] hypothesis:

Hypothesis: "Input validation missing from /api/upload" [INFERRED]

Action: Read src/app/api/upload/route.ts
Found: Zod schema validation at line 18, applied before file processing.
Result: "Input validation exists at api/upload/route.ts:18" [VERIFIED]
Hypothesis was wrong.
Hypothesis: "Database queries might not be parameterized" [INFERRED]

Action: Search for raw SQL patterns (string interpolation in queries)
Grep: `db.query(\`.*\${`, `db.execute(\`.*\${`
Read: src/lib/db.ts, src/repositories/*.ts
Found: All queries use prepared statements. One raw query at analytics.ts:89.
Result: "Raw SQL injection risk at src/analytics.ts:89" [VERIFIED]

Only after Pass 2 are claims safe to report.

Quick Checklist

Before making any codebase claim, confirm:

  • Have I read the actual file, not just grep results?
  • Is the code in a live path (not commented out, not test-only)?
  • For absence claims: did I try 3+ alternative search terms?
  • For absence claims: did I manually check the likely locations?
  • For behavior claims: did I read the full function, not just the name?
  • Does my claim include a file:line citation?
  • Is my marker accurate: [VERIFIED] / [INFERRED] / [UNCERTAIN]?

Integration Points

claim-verification.md — the rule file provides guidelines; this skill provides the action protocol.

Agents that must use this skill:

  • code-reviewer: before asserting "the code lacks X"
  • security-reviewer: before asserting "vulnerability exists/doesn't exist"
  • architect: before asserting "pattern is/isn't used"
  • sleuth: before asserting "root cause is X"
  • coroner: before asserting "same bug exists elsewhere"

Hook integration (optional): After every Grep tool result, the epistemic-reminder hook injects a warning. That warning is this skill's Level 1 trigger — search results are [INFERRED], not [VERIFIED].

Examples: Wrong vs Right

Existence claim — WRONG

Grep found "authMiddleware" in 3 files.
Claim: "Auth middleware is applied to all routes."

Existence claim — RIGHT

Grep found "authMiddleware" in 3 files.
Read server.ts: middleware applied only to /api/* routes, not /public/*.
Read admin/router.ts: authMiddleware missing from /admin/health endpoint.
Claim: "Auth middleware applied to /api/* (server.ts:23) but missing from
/admin/health (admin/router.ts:8)." [VERIFIED]

Absence claim — WRONG

Grep "rate limit" returned no results.
Claim: "No rate limiting implemented."

Absence claim — RIGHT

Grep "rate limit" → no results.
Grep "rateLimit", "throttle", "limiter", "requestLimit" → no results.
Read: src/middleware/index.ts, server.ts, next.config.js.
Confirmed none contain request throttling logic.
Claim: "No rate limiting found after searching 6 term variants and
reading 3 likely files." [VERIFIED ABSENT]

© vibeeval, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/factcheck-guard of vibeeval/vibecosystem.

Open the folder on GitHubat commit 3b763b1

Compare with similar skills

Factcheck Guard next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Factcheck Guard compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Factcheck Guard this skillvibeeval/vibecosystem532—~2.2kAutomated safety check: PassMIT
Perplexity Web Searchdavila7/claude-code-templates33k11 repos~3.5kAutomated safety check: NotesMIT
Citation Verification GuideGalaxy-Dawn/claude-scholar5.7k2 repos~1.9kAutomated safety check: PassMIT
Article Fact Checkerdigoal/blog8.6k—~939Automated safety check: PassGPL-2.0
Deep Research Agent TeamImbad0202/academic-research-skills51k—~13kAutomated safety check: PassCustom licence
Docs Grounding Verifiermicrosoft/apm4k—~1.9kAutomated safety check: PassMIT

Similar skills

  • Perplexity Web Search

    davila7/claude-code-templates

    Runs web-grounded searches through Perplexity's Sonar models over OpenRouter for current events, recent literature and cited facts beyond the model's training cutoff.

    33k GitHub starsUsed in 11 repos~3.5k tokens
    Research & ScienceAuto-check: notes

  • Citation Verification Guide

    Galaxy-Dawn/claude-scholar

    Reference guidance for checking every citation in academic writing against canonical sources such as DOI, arXiv, CrossRef and Semantic Scholar, to catch fake or wrong references.

    5.7k GitHub starsUsed in 2 repos~1.9k tokens
    Research & ScienceAuto-check passed

  • Article Fact Checker

    digoal/blog

    三层审查模型,逐段逐句验证文章真伪、证据链与逻辑结构。Use when the user asks to fact-check, verify, audit, or evaluate the credibility of an article, essay, report, opinion piece, social-media post, or any written claim —…

    8.6k GitHub stars~939 tokensUpdated yesterday
    Research & ScienceAuto-check passed

  • Deep Research Agent Team

    Imbad0202/academic-research-skills

    Runs a 13-agent pipeline for rigorous academic research, from forming the question through systematic search, synthesis, bias checks and an APA 7.0 report.

    51k GitHub stars~13k tokensUpdated yesterday
    Research & ScienceAuto-check passed

  • Docs Grounding Verifier

    microsoft/apm

    Official

    A skill your agent uses to verify CLAIM-LEVEL grounding of a documentation page (or set of pages) against the source code.

    4k GitHub stars~1.9k tokensUpdated today
    Research & ScienceAuto-check passed

  • Fact Checking

    bradygaster/squad

    Review and validate claims using counter-hypothesis testing.

    3.3k GitHub stars~503 tokensUpdated today
    Research & ScienceAuto-check passed

More from vibeeval/vibecosystem

All 12 skills in this repo

  • Agent Benchmark

    vibeeval/vibecosystem

    Framework for measuring and tracking agent response quality over time.

    532 GitHub stars~2.9k tokensUpdated 2 mo ago
    Auto-check passed

  • Differential Review

    vibeeval/vibecosystem

    Security-focused differential code review with blast radius analysis, risk-adaptive depth (DEEP/FOCUSED/SURGICAL), git history correlation, and structured finding format.

    532 GitHub stars~1.6k tokensUpdated 2 mo ago
    Auto-check passed

  • Fp Check

    vibeeval/vibecosystem

    Systematic false positive verification for security findings.

    532 GitHub stars~1.6k tokensUpdated 2 mo ago
    Auto-check passed

  • N8n Workflows

    vibeeval/vibecosystem

    n8n otomasyon workflow'lari. An agent skill from vibeeval/vibecosystem.

    532 GitHub stars~3.3k tokensUpdated 2 mo ago
    Auto-check passed

  • Notepad System

    vibeeval/vibecosystem

    A skill your agent uses when context compression is imminent, when resuming a session, or when preserving critical decisions across long tasks.

    532 GitHub stars~1.7k tokensUpdated 2 mo ago
    Auto-check passed

  • Property Based Testing

    vibeeval/vibecosystem

    Property-based testing (PBT) patterns with fast-check (JS/TS), Hypothesis (Python), and gopter (Go).

    532 GitHub stars~2k tokensUpdated 2 mo ago
    Auto-check passed

Categories

Questions about Factcheck Guard

What does Factcheck Guard do?

A skill your agent uses when making any factual claim about the codebase — existence, absence, or behavior. Factcheck Guard is an agent skill from vibeeval/vibecosystem. Use this skill when making any factual claim about the codebase — existence, absence, or behavior.

When should I use Factcheck Guard?

Factcheck Guard fits situations like: making any factual claim about the codebase — existence; tasks that involve Fact-checking and source verification.

How do I install Factcheck Guard in Claude Code?

Run `npx skills add vibeeval/vibecosystem --skill factcheck-guard -a claude-code`. Or copy the skill folder (skills/factcheck-guard in vibeeval/vibecosystem) into .claude/skills/factcheck-guard in your project. Claude Code loads it when a task matches its description.

How do I install Factcheck Guard in Codex?

Run `npx skills add vibeeval/vibecosystem --skill factcheck-guard -a codex`. Or copy the skill folder (skills/factcheck-guard in vibeeval/vibecosystem) into .agents/skills/factcheck-guard in your project. Codex loads it when a task matches its description.

Can I use Factcheck Guard in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add vibeeval/vibecosystem --skill factcheck-guard -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/factcheck-guard, .gemini/skills/factcheck-guard, .github/skills/factcheck-guard and .opencode/skills/factcheck-guard in your project.

What does Factcheck Guard need to run?

SKILL.md names no scripts, command-line tools or credentials: Factcheck Guard is instructions for the agent only.

Does Factcheck Guard access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Factcheck Guard safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Factcheck Guard use?

Factcheck Guard is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Factcheck Guard use?

About 2.2k tokens (SKILL.md is roughly 8.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Factcheck Guard?

Skills that share tags, products or a category with Factcheck Guard: Perplexity Web Search (davila7/claude-code-templates, 33k stars), Citation Verification Guide (Galaxy-Dawn/claude-scholar, 5.7k stars), Article Fact Checker (digoal/blog, 8.6k stars) and Deep Research Agent Team (Imbad0202/academic-research-skills, 51k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Factcheck Guard?

vibeeval (a GitHub user) maintains it in vibeeval/vibecosystem, which has 532 GitHub stars. The repository holds 12 skills in this directory. The repository was last updated on August 8, 2026.

Source: vibeeval/vibecosystem on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.