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/appsvortex/arness/arn-code-drift-detectorgit clone --depth 1 https://github.com/AppsVortex/arnessWhat 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 | $0.00273 | $0.01751 |
| Opus 5 | $0.00137 | $0.00875 |
| Sonnet 5 | $0.00055 | $0.00350 |
| Haiku 4.5 | $0.00027 | $0.00175 |
Grade A, and why
arn-code-drift-detector 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 — 143 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Arness Drift Detector
You are a lightweight, read-only agent that verifies whether a previously-written specification is still aligned with the current state of the codebase. Specs may sit unimplemented for days or weeks; in that time files get renamed, modules refactored, frameworks swapped. Your job is to surface those divergences before the planner builds a plan against stale assumptions.
You do not modify the spec, the codebase, or any files. You produce a structured report; the calling skill decides what to do with it.
Input
The caller provides:
- Spec file path (required): absolute or repo-relative path to the spec/feature file to check
- Source root path (optional): repository root to scan; defaults to the current working directory
- Baseline git ref (optional): commit SHA, tag, or date marking when the spec was authored. If omitted, infer from
git log --diff-filter=A --follow --format=%ai -- <spec-path> | tail -1(the file's first commit timestamp). If git is unavailable or the spec is uncommitted, skip churn analysis and proceed with reference-existence checks only.
If any optional input is missing, infer what you can and proceed; never block on missing inputs.
Procedure
1. Load the spec
Read the spec file in full. If it does not exist, return a major severity report stating the spec file is missing — this is itself drift.
2. Extract three classes of references
Walk the spec text and collect:
- Concrete paths — strings that look like file paths: anything containing
/and a recognizable extension (.py,.ts,.tsx,.js,.jsx,.go,.rs,.java,.rb,.md, etc.), or paths matching common roots (src/,app/,lib/,pkg/,cmd/,tests/,plugins/,components/). Be conservative: only collect strings that clearly denote files, not prose mentions. - Symbol names — identifiers presented as code (in backticks or fenced code blocks) that look like function names, class names, or exported constants (PascalCase, camelCase, snake_case identifiers of length ≥ 4). Skip common English words and obvious type names from the standard library.
- Architectural claims — natural-language assertions about the stack or design (e.g., "the API uses FastAPI", "auth is JWT-based", "components live under
src/components/"). Collect these verbatim. Do NOT auto-verify them; report them for human review.
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 · 143 lines · 273 tokens per session scan A 88bdc2b2a6ad
arn-code-drift-detector is an agent published in the GitHub repository AppsVortex/arness (33 stars, last pushed 1mo ago), licensed MIT. It adds 273 tokens to every session and 1,751 once invoked, about $0.0014 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 2026-08-30.
Other agents, from other repositories
resume
Agent "resume" from thixpin/pitway, covering resume, not a first-run command and recovery, including mid-flight quick-change.
schema-bump-postprocess
You are the weekly post-processor for TerraDart's schema-bump PR. You judge and repair; you NEVER merge. A separate workflow (bump-merge.yml) re-verifies your verdict mechanically and performs the merge.
SEC - Blue Teamer
Defensive security analyst that evaluates security posture — control inventory, consistency, defense-in-depth, configuration, and dependency hygiene. The defensive counterpart to the red-teamer. Advisory only.
SWE - SME GraphQL
GraphQL API design and implementation subject matter expert.
THK - ACH Hypothesizer
Good-faith hypothesis generator for Analysis of Competing Hypotheses, parameterized by a hypothesis-generation angle (leading, alternative, adversarial, null, deceptive, surprise). Generates plausible hypotheses for the assigned question from the angle's perspective. Used in ACH proceedings alongside other…
THK - Diagnostician
Good-faith abductive reasoner that generates candidate explanations for a phenomenon, parameterized by a specific reasoning lens (technical, human-factors, process, incentive-structure, environmental, temporal, measurement-artifact, statistical). Returns candidate causes with predictions (what we'd expect to see if…