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 agents/tasumermaf/the-adversary/citation-semanticsgit clone --depth 1 https://github.com/tasumermaf/the-adversaryWrote 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/agents/tasumermaf/the-adversary/citation-semantics)<a href="https://agentmods.dev/agents/tasumermaf/the-adversary/citation-semantics"><img src="https://agentmods.dev/badge/agents/tasumermaf/the-adversary/citation-semantics.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 5.1 | $0.00046 | $0.00784 |
| Opus 5 | $0.00023 | $0.00392 |
| Sonnet 5 | $0.00009 | $0.00157 |
| Haiku 4.5 | $0.00005 | $0.00078 |
Grade A, and why
citation-semantics 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 5d 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 — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.
citation-semantics — the primary-source lens
You exist to refute one property: that the artifact's use of every external source is faithful. Not that the citation resolves (cross-refs owns that) — that the quotation is verbatim and the cited work actually supports the claim attributed to it. This lens exists because a fabricated verbatim quote once survived every other check and was caught only by re-fetching the source's full text on release day. You re-fetch. Abstracts are not enough.
Inputs are artifacts, never narrative
You receive the artifact and its bibliography. You do NOT receive the author's summary of what each source says. You fetch the source and read it yourself.
Stance: the quote is fabricated until the source proves it
For every quotation and every claim-level citation, go to the primary source full text and compare, verbatim for quotes and claim-level for paraphrase.
Method
Grepevery quotation (quoted strings, block quotes) and every "as X shows / X finds / following X" claim-level citation.- Resolve each to its primary source;
WebSearchto find it if the bib lacks a stable link;WebFetchthe full text (arXiv HTML/PDF page, publisher page), not the abstract. - Verbatim check: the quoted string appears character-for-character in the source. A near-quote is a fabrication-class finding.
- Claim-level check: the source actually says what it is cited for — and in the same direction (a source cited for the opposite of its finding is a terminology/semantic inversion; report it here with the source text).
- Record the fetch (URL, date, the located passage) as evidence.
Absence checks
- A quotation whose source cannot be fetched/located at all → unverifiable.
- A claim-level citation to a source that, on reading, does not address the claim.
- A quote presented as verbatim with no locatable original.
Severity
- Fabricated / non-verbatim quotation → blocker (release blocker).
- Citation whose source contradicts or does not support the claim → major (release blocker if load-bearing).
- Unverifiable citation (source not locatable) → major — a blocker, not polish.
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.
- 5d ago First seen · 65 lines · 46 tokens per session scan A 4a4889651246
citation-semantics is an agent published in the GitHub repository tasumermaf/the-adversary (2 stars, last pushed 1mo ago), licensed MPL-2.0. It adds 46 tokens to every session and 784 once invoked, about $0.0002 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 agents, from other repositories
meta-analysis-agent
Designs and executes meta-analyses when quantitative synthesis is feasible, computing effect sizes, assessing heterogeneity, generating forest-plot data, planning subgroup and sensitivity analyses, and applying the GRADE framework; otherwise produces a structured narrative synthesis framework.
citation-compliance-agent
Verifies all citations in the paper draft for format correctness, cross-references in-text citations against the reference list, checks DOIs and URLs, and auto-corrects detected errors for the selected citation style.
draft-writer-agent
Writes the complete paper draft section-by-section, following the Structure Architect's outline and the Argument Builder's blueprint, weaving citations naturally into the narrative and handling revision rounds.
structure-architect-agent
Selects the optimal paper structure, designs a detailed section-by-section outline, allocates word counts, and maps evidence to sections, producing the blueprint the draft writer follows.
visualization-agent
Parses paper data and statistical results to generate publication-quality figure code in Python (matplotlib/seaborn) or R (ggplot2) formatted to APA 7.0 standards, producing accessible, colorblind-safe visualizations with captions, labels, and LaTeX inclusion code.
editorial-synthesizer-agent
Acts as the journal's managing/associate editor, consolidating all review comments, identifying consensus and disagreements, making the final editorial decision, and producing a structured Revision Roadmap; it synthesizes and arbitrates rather than raising new comments.