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 skills add sergeyizmailov/knowledge-delta-skills --skill knowledge-delta-skill-architectgit clone --depth 1 https://github.com/sergeyizmailov/knowledge-delta-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/skills/sergeyizmailov/knowledge-delta-skills/knowledge-delta-skill-architect)<a href="https://agentmods.dev/skills/sergeyizmailov/knowledge-delta-skills/knowledge-delta-skill-architect"><img src="https://agentmods.dev/badge/skills/sergeyizmailov/knowledge-delta-skills/knowledge-delta-skill-architect/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sergeyizmailov/knowledge-delta-skills/knowledge-delta-skill-architect"><img src="https://agentmods.dev/badge/skills/sergeyizmailov/knowledge-delta-skills/knowledge-delta-skill-architect.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00091 | $0.03397 |
| Opus 5.5 | $0.00036 | $0.01359 |
| Sonnet 5.5 | $0.00018 | $0.00679 |
| Haiku 4.5 | $0.00009 | $0.00340 |
Grade A, and why
knowledge-delta-skill-architect 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 1mo 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 — 230 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Knowledge-Delta Skill Architect
Reviewed 2026-09-09. Method unmeasured — no baseline was run for this skill itself (the
Baseline: stamp §"Scope, then baseline" mandates is therefore absent by disclosure, not
oversight). Runtime numbers in 02 §0 were verified 2026-08-28 / 09-08.
01 02 03 = the numbered files under references/; read triggers in the table at the end.
Sources and their limits: 03.
The method here is model- and vendor-independent. What is not — numbers, runtime behavior — is confined
to 02: §0 for the numbers, inline flags wherever behavior is runtime-specific. When a spec or a
runtime changes, refresh 02.
Spend context only where the model's behavior changes.
Write for the target model, not as a human tutorial. Omit orientation, motivation, definitions, examples and spelled-out reasoning unless the baseline showed they prevent a failure; the gap is sometimes a missing definition. Keep decision rules, conditions, thresholds, discriminators, and the rationale that makes a rule generalize. Your default register is documentation; override it on purpose.
Classes
| Class | What you saw | Content | Budget |
|---|---|---|---|
| [W] | States it wrong, confidently, and never asks | Overwrite the prior | Highest value per token |
| [K] | Doesn't know, hedges, or invents plausibly | The fact | Scale to blast radius |
| [A] | Knows it, skips it under task pressure | One imperative line | Only if you saw the skip |
| [O] | Solves it acceptably, never considers the stronger option | That option + when it wins | High — most lore lands here |
| [R] | Gets it right unprompted, and nothing better exists | — | Delete |
[R] is a correctness bug, not waste: irrelevant context degrades reasoning, and content topically
adjacent to the task degrades it most — telling the model to ignore it barely helps. A redundant
paragraph inside your own domain is the most expensive token you can spend [inference]. Numbers: 03.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 1mo ago Changed · +4 lines e5f8200287c5
- 1mo ago Changed · +19 lines 7483e610015a
- 1mo ago First seen · 207 lines · 91 tokens per session scan A 8434fea86f7b
knowledge-delta-skill-architect is a skill published in the GitHub repository sergeyizmailov/knowledge-delta-skills (3 stars, last pushed 1mo ago), licensed MIT. It adds 91 tokens to every session and 3,397 once invoked, about $0.0004 per session on Opus 5.5. 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 skills, from other repositories
firebase-ai
Use when setting up firebaseai, generating text/chat with Gemini, streaming AI output, building multimodal prompts, or handling AI errors.
karpathy-llm-wiki
Use when building or maintaining a personal LLM-powered knowledge base. Triggers: ingesting sources into a wiki, querying wiki knowledge, linting wiki quality, 'add to wiki', 'what do I know about', or any mention of 'LLM wiki' or 'Karpathy wiki'.
standalone-python-scripts
Skill "standalone-python-scripts" from iloveitaly/llm-ide-rules, covering standalone python scripts, /// script, requires-python = ">=3.13", dependencies = [] and ///.
context-optimization
Optimize effective context capacity through compression, masking, caching, and partitioning strategies. Use when the user asks to reduce token costs, improve context efficiency, implement KV-cache optimization, or extend context-window effectiveness.
rag-implementation
Build Retrieval-Augmented Generation (RAG) systems for LLM applications with vector databases and semantic search. Use when implementing knowledge-grounded AI, building document Q&A systems, or integrating LLMs with external knowledge bases.
setup
Configure the Qdrant Power after installation. Use this skill for missing uvx, missing environment variables, unapproved environment variables, unavailable Qdrant tools, "Failed to connect" errors, and setup requests.