cite-or-refuse

cite-or-refuse is a skill for Claude Code, Codex from ArdurAI/ardur-agent. It costs 24 tokens per session (773 once invoked), scanned A, original, Apache-2.0.

A skill for answering questions using retrieved material from a private workspace or user-provided collection.

In plain words
What is it for?
It helps produce cited answers grounded in stored documents, with each factual claim linked to a retrieved source span.
Why use it?
It prevents unsupported claims when the needed information is not available. If retrieval returns nothing usable, it requires the agent to refuse instead of guessing.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit It helps produce cited answers grounded in stored documents, with each factual claim linked to a retrieved source span.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/ardurai/ardur-agent/cite-or-refuse
Install

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.

Any agent
npx skills add ArdurAI/ardur-agent --skill cite-or-refuse
Clone the repo
git clone --depth 1 https://github.com/ArdurAI/ardur-agent

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.

agentmods badge for cite-or-refuse

README.md
[![agentmods](https://agentmods.dev/badge/skills/ardurai/ardur-agent/cite-or-refuse/github.svg)](https://agentmods.dev/skills/ardurai/ardur-agent/cite-or-refuse)
Your own site
<a href="https://agentmods.dev/skills/ardurai/ardur-agent/cite-or-refuse"><img src="https://agentmods.dev/badge/skills/ardurai/ardur-agent/cite-or-refuse/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.

agentmods 80×15 button for cite-or-refuse

Your own site · 80×15
<a href="https://agentmods.dev/skills/ardurai/ardur-agent/cite-or-refuse"><img src="https://agentmods.dev/badge/skills/ardurai/ardur-agent/cite-or-refuse.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 24 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 773 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. 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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00024 $0.00773
Opus 5 $0.00012 $0.00387
Sonnet 5 $0.00005 $0.00155
Haiku 4.5 $0.00002 $0.00077

Measured 8d ago against content hash 812f236e45ab, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

cite-or-refuse 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 8d 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.

examples/skills/cite-or-refuse/SKILL.md · 83 lines

How it starts

The opening of the file, as written. The whole thing — 83 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Cite-or-refuse grounded mode

Use this skill whenever the user requests an answer that depends on private, workspace, or user-corpus knowledge rather than general reasoning. The answer is allowed only when it is grounded in retrieved spans from the user corpus.

Mandatory preflight

  1. Query the configured user-corpus retriever before drafting the answer.
  2. Treat the retrieval result as empty when there are zero spans, every span has an unusable body, or the retrieval layer reports an authorization/error state that prevents reading the corpus.
  3. If retrieval is empty, do not answer from memory or model prior knowledge. Refuse with the template below and include a grounding receipt showing status: refused_empty_retrieval.
  4. If retrieval returns spans, use only those spans for corpus-dependent claims. General-language glue is fine, but factual claims about the corpus require a citation.

Citation rules

  • Cite each corpus-dependent sentence with one or more source span markers.
  • Source markers use the stable span id form [S<n>], where each marker maps to one retrieved source span in the source list.
  • Never cite a document as a whole when the retriever returned narrower spans.
  • Do not merge nearby spans unless the retrieval layer explicitly returned the merged span.
  • If a requested detail is not present in the retrieved spans, say that the retrieved corpus does not support that detail instead of guessing.

Required response shape

When grounded spans are available, respond in this order:

  1. Answer — concise answer with inline [S<n>] citations on every corpus-dependent claim.
  2. Sources — one bullet per cited span, including span id, document/source title or URI, byte/line/time offsets when available, and a short quoted excerpt. Keep excerpts minimal and do not invent offsets.
  3. Grounding receipt — a compact machine-readable block following @./receipt-schema.md.

When retrieval is empty, use this refusal template:

Read the full file on GitHub · 83 lines

Files

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.

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.

  1. 8d ago First seen · 83 lines · 24 tokens per session scan A 812f236e45ab

Subscribe to this mod's changes

cite-or-refuse is a skill published in the GitHub repository ArdurAI/ardur-agent (2 stars, last pushed 4d ago), licensed Apache-2.0. It adds 24 tokens to every session and 773 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 2026-08-31.

Related

Other skills, from other repositories