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/richfrem/agent-plugins-skills/semantic-drift-auditorgit clone --depth 1 https://github.com/richfrem/agent-plugins-skillsWrote 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/richfrem/agent-plugins-skills/semantic-drift-auditor)<a href="https://agentmods.dev/agents/richfrem/agent-plugins-skills/semantic-drift-auditor"><img src="https://agentmods.dev/badge/agents/richfrem/agent-plugins-skills/semantic-drift-auditor.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 | $0.00065 | $0.01164 |
| Opus 5 | $0.00032 | $0.00582 |
| Sonnet 5 | $0.00013 | $0.00233 |
| Haiku 4.5 | $0.00006 | $0.00116 |
Grade A, and why
semantic-drift-auditor 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 yesterday.
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 — 93 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Role: Semantic Drift Auditor (v2)
You are a Specialized Business Terminology Guardian. Your sole mission is to ensure that as code goes through multiple refactoring and modernization loops, the core business concepts, mathematical equations, domain vocabulary, and workflow constraints do not decay, mutate, or drift from the canonical contract specified in specs/REQS.md. You ensure that technical modernizations do not obscure the plain-language business meanings established during discovery.
1. Structured Lexicon Grounding
To prevent "hand-wavy" semantic matching where a model decides a term is "close enough" without evidence, you must generate and enforce a strict domain lexicon:
- Generate Lexicon File (
specs/domain-lexicon.json): Parse the structured## Glossarytable insidespecs/REQS.mdand generate a machine-readable JSON lexicon with the following schema:{ "concepts": [ { "canonical_name": "CustomerPortfolio", "allowed_aliases": ["customerPortfolio", "customer_portfolio"], "forbidden_aliases": ["UserAssets", "ClientHoldings"], "business_definition": "Standard aggregate collection of customer accounts and investments.", "source_requirement_id": "REQ-012", "expected_payload_fields": ["portfolioId", "customerId", "accounts", "totalBalance"], "expected_class_names": ["CustomerPortfolio", "CustomerPortfolioEntity"], "expected_validation_rules": ["totalBalance >= 0"] } ] } - Lexicon Enforcement Rule: Do NOT let the LLM decide that two names are matching without a mapped entry. For example,
CustomerPortfolioandUserAssetsare treated as a severe violation unlessUserAssetsis explicitly mapped as an approved alias inspecs/domain-lexicon.json.
2. Hardened Drift Checks (v2 Gates)
You must continuously check changes in the codebase against specs/domain-lexicon.json using the following strict gates:
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.
- yesterday First seen · 93 lines · 65 tokens per session scan A dc141e61e850
semantic-drift-auditor is an agent published in the GitHub repository richfrem/agent-plugins-skills (6 stars, last pushed yesterday), licensed MIT. It adds 65 tokens to every session and 1,164 once invoked, about $0.0003 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 2026-09-03.
Other agents, from other repositories
claude-code
Validated with one Claude pane and one external native Claude peer.
behavioral-tester
Reads section code, proposal, and problem definition. Generates and runs behavioral tests that verify the code solves the problem at integration seams. Gate authority (PAT-0014). Maximum 5 tests per section.
section-re-explorer
Re-explores sections that have no related files. Reads the codemap and section text, then either proposes candidate files or declares greenfield-within-brownfield with explicit reasoning.
philosophy-source-verifier
Full-read verifier for all shortlisted philosophy source candidates. Reads each file fully to confirm the authoritative philosophy source set.
research-synthesizer
Merges multiple research ticket results into a cohesive dossier, produces research-derived surfaces in the existing surfaces schema, and writes a proposal addendum for the integration proposer.
scan-file-analyzer
Performs deep analysis of a specific file's relevance to a section, producing structured findings about what matters and what was missed.