verify-against-code

A checking rule for statements about what code or a product does. It requires reading the relevant implementation or other direct evidence before presenting such claims as facts.

In plain words
What is it for?
Use it when answering whether a product or code path supports a feature, and when writing outward-facing documentation, commit messages, product descriptions, or reports about behavior.
Why use it?
It reduces the chance of publishing outdated or incorrect descriptions based on memory, assumptions, or plans that were never implemented.

Skill for Claude CodeCodex

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.

agentmods
npx agentmods add skills/fergius-engineering/instincts/verify-against-code
Any agent
npx skills add Fergius-Engineering/instincts --skill verify-against-code
Clone the repo
git clone --depth 1 https://github.com/Fergius-Engineering/instincts

Made for: Claude Code, Codex.

Per session 40 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 475 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00040 $0.00475
Opus 5 $0.00020 $0.00237
Sonnet 5 $0.00008 $0.00095
Haiku 4.5 $0.00004 $0.00047

Measured 2d ago against content hash d74f2be9f7d5, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

verify-against-code 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 2d 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.

skills/verify-against-code/SKILL.md · 34 lines

What it actually says

The rule

Before you state a fact about what the code or product does, go read the thing that proves it. Your memory, the user's framing, a plan that says "done", a table in the docs — all of these go stale, and stating one as fact when it's wrong costs more than the thirty seconds it takes to check.

Fires when

Answering "does it do X?", writing a doc or README line, writing a commit or PR body, shipping anything outward-facing (store copy, a caption, a before/after), agreeing with a claim the user just made about the code.

How to apply

For each claim, find the concrete artifact that proves it — the symbol, the field, the default value, the line — and read it. Cite what you read as the evidence, briefly.

"It doesn't actually do that" and "only partially" are valid, valuable answers. Say so plainly and surface the conflict instead of going along with the assumption.

Hold this bar even when nobody asked you to check. Especially when nobody asked.

Worked example

A user asks: "Our API trims whitespace from usernames before saving, right?" You remember writing something like that. The cheap move is "yes." The right move: open the save path and read it. You find username.toLowerCase() but no .trim(). The honest answer is "It lowercases, but it does not trim — leading spaces are saved as-is." That gap is exactly the bug the user was about to assume away. The thirty-second read changed a confident wrong answer into a real finding.

Red flags

Thought Reality
"I'm pretty sure it does that" Sure isn't read. Open the file.
"The plan says it's done" Plans go stale. The code is the truth.
"The user said it works that way" The user can be wrong about their own system. Check.
"It's just a doc line, not code" A wrong doc line is a wrong claim shipped to every reader.
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. 2d ago First seen · 34 lines · 40 tokens per session scan A d74f2be9f7d5

Subscribe to this mod's changes

verify-against-code is a skill published in the GitHub repository Fergius-Engineering/instincts (2 stars, last pushed 7d ago), licensed MIT. It adds 40 tokens to every session and 475 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.

Related

Other skills, from other repositories

rulesync

Generates and syncs AI rule configuration files (.cursorrules, CLAUDE.md, copilot-instructions.md) across 20+ coding tools from a single source. Use when syncing AI rules, running rulesync commands, importing or generating rule files, or managing shared AI coding configurations.

dyoshikawa/rulesync · 64 tokens

establishing-project-context

Use when the user asks to establish shared project language, or project work exposes a conflicting, renamed, or deprecated domain term that needs active semantic modeling. Routine small tasks stay on the fast path.

GanyuanRan/Aegis · 45 tokens

autoprompt

Explicit-only useful-first orchestration. Invoke /autoprompt to turn a mission into one executable roadmap, build dependency-safe lanes, and verify the result with independent reviewers. Never infer invocation from ordinary requests. Never resume from leftover artifacts without an explicit resume instruction.

Spielewoy/autoprompt-skill · 56 tokens

office-hours

YC Office Hours — two modes. Startup mode: six forcing questions that expose demand reality, status quo, desperate specificity, narrowest wedge, observation, and future-fit. Builder mode: design thinking brainstorming for side projects, hackathons, learning, and open source. Saves a design doc. Use when asked to…

GCWing/BitFun · 168 tokens

memstack-business-gdpr

Use this skill when the user says 'GDPR', 'data protection', 'privacy compliance', 'DPA', 'DSAR', 'data subject request', 'cookie consent', 'privacy audit', 'CCPA', or asks 'do I need GDPR for this repo'. Scans the repository to detect what personal data is collected, classifies sensitivity, determines whether GDPR…

cwinvestments/memstack · 121 tokens

echo

Use when the user references past sessions, asks 'what did we do', 'do you remember', 'last session', 'recall', or 'continue from'.

cwinvestments/memstack · 35 tokens