kernel-verify

A procedure for checking whether answering from shortened context produces the same conclusion as answering from the full context. The comparison is called answer invariance.

In plain words
What is it for?
Use it by answering one question separately from the full and reduced contexts, then reporting whether the answers match and identifying any missing information if they do not.
Why use it?
It reveals whether context compression or projection removed information that changes the answer, without using another API or incurring extra API costs.

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/pinperepette/context-kernel/kernel-verify
Any agent
npx skills add Pinperepette/context-kernel --skill kernel-verify
Clone the repo
git clone --depth 1 https://github.com/Pinperepette/context-kernel

Made for: Claude Code, Codex.

Per session 62 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 665 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 $0.00062 $0.00665
Opus 5 $0.00031 $0.00332
Sonnet 5 $0.00012 $0.00133
Haiku 4.5 $0.00006 $0.00067

Measured 2d ago against content hash 3757017fe1ed, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

kernel-verify 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.

claude-context-kernel/skills/kernel-verify/SKILL.md · 53 lines

What it actually says

kernel-verify — answer-invariance in-sessione (zero chiavi)

Serve a rispondere a una sola domanda: la versione compressa/proiettata del contesto porta alla stessa risposta di quella completa? Cioe' A(x) = A(pi(x)).

Questo controllo non usa API ne' chiavi: lo esegui tu, Claude, dentro la sessione — quindi pesa solo sull'abbonamento gia' attivo.

Input

L'utente ti fornisce (o indica dove trovarli):

  1. un task/domanda Q;
  2. il contesto completo x;
  3. il contesto ridotto pi(x) (output di compress.py, dello slicer, o di un proiettore span(Q)).

Procedura (falla in silenzio, poi riporta solo il verdetto)

  1. Rispondi a Q usando solo x. Chiama questa A(x).
  2. Rispondi a Q usando solo pi(x). Chiama questa A(pi(x)). Trattale come due contesti separati: non lasciare che x "riempia i buchi" di pi(x).
  3. Confronta A(x) e A(pi(x)) nel merito (stessa conclusione, stessi fatti rilevanti). Ignora differenze di forma, lunghezza o parole.

Output

Riporta in modo compatto:

  • INVARIANTE: si/no
  • se no: quale informazione presente in x e' andata persa in pi(x) e ha cambiato la risposta (indica l'unita'/riga mancante).
  • una riga di verdetto (es. "stessa risposta: 30s di timeout" oppure "persa: il limite upload del Pro, ora la risposta e' incompleta").

A cosa serve

E' il gate che trasforma la compressione euristica (conservativa, non provabile) in una misura verificata per quel task. Usalo per:

  • tarare quanto puoi comprimere prima che l'invarianza si rompa (il "ginocchio" della curva rate-distortion di span_rd.py);
  • validare a campione l'output del compressore su casi delicati.

Nota: lo slicer del codice (kernel-slice) e' gia' answer-preserving per costruzione — su quello questo controllo e' ridondante. Serve soprattutto sul testo/documenti, dove la proiezione e' empirica.

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 · 53 lines · 62 tokens per session scan A 3757017fe1ed

Subscribe to this mod's changes

kernel-verify is a skill published in the GitHub repository Pinperepette/context-kernel (24 stars, last pushed 1mo ago), licensed MIT. It adds 62 tokens to every session and 665 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 2026-08-30.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

babysit-pr

Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…

openai/codex · 114 tokens

imagegen

Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…

openai/codex · 113 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens

next-cache-components-optimizer

Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…

vercel/next.js · 170 tokens