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 aka-kika/akakika-skills --skill butterfly-effectgit clone --depth 1 https://github.com/aka-kika/akakika-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/aka-kika/akakika-skills/butterfly-effect)<a href="https://agentmods.dev/skills/aka-kika/akakika-skills/butterfly-effect"><img src="https://agentmods.dev/badge/skills/aka-kika/akakika-skills/butterfly-effect/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/aka-kika/akakika-skills/butterfly-effect"><img src="https://agentmods.dev/badge/skills/aka-kika/akakika-skills/butterfly-effect.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00172 | $0.01356 |
| Opus 5 | $0.00086 | $0.00678 |
| Sonnet 5 | $0.00034 | $0.00271 |
| Haiku 4.5 | $0.00017 | $0.00136 |
Grade A, and why
butterfly-effect 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 12d 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 — 84 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Butterfly Effect
A session summary tells you what happened. A butterfly trace tells you which moments were load-bearing — the two to four points where, if something had gone differently, the entire outcome would not exist. Most of a long session is noise: dead ends, routine steps, things that would have happened in any version of the session. The skill's job is filtering, not recording.
The core move is the counterfactual test, applied backward from the outcome:
If not this — then this would not happen.
An event is a butterfly point only if removing it breaks the chain to the final result. "We ran the tests" is not a butterfly point — tests would have run in any version of this session. "Test 3 failed with a timezone error, which is what made us look at the date parsing at all" is a butterfly point: no failure, no looking, no fix.
When this runs
At session close, or when the user invokes it mid-session on the work so far. If the session produced no meaningful outcome (pure exploration, no decision, nothing shipped), say so and skip the trace — an empty trace is more honest than a forced one.
The procedure
Work backward, never forward. Forward walks reproduce the session's own noise; backward walks force every step to justify its causal connection to the end.
-
Name the outcome. One sentence. The thing that exists now that did not exist at session start: the fix, the decision, the renamed project, the abandoned approach. If there are multiple independent outcomes, trace each separately.
-
Ask what the outcome directly depended on. Not what preceded it — what it required. Usually one or two things.
-
Recurse. For each dependency, ask what it required. Keep going until you hit session start or an external input (something the user brought in, an error the world produced, a message that arrived).
-
Apply the counterfactual test to every node. Delete the node mentally. Does the outcome still plausibly happen, maybe later or by another route? Then it is not a butterfly point — cut it. Only keep nodes whose removal genuinely severs the chain.
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.
- 12d ago First seen · 84 lines · 172 tokens per session scan A 7aea7b4ae81b
butterfly-effect is a skill published in the GitHub repository aka-kika/akakika-skills (10 stars, last pushed 2d ago), licensed MIT. It adds 172 tokens to every session and 1,356 once invoked, about $0.0009 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.
Other skills, from other repositories
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'.
init
Install or update OwnMem in the current repository. Use when the user asks to set up OwnMem, add local project memory for coding agents, or refresh an existing OwnMem installation after a version bump.
memorix-sessions
Use when resuming work, preparing handoff context, binding an HTTP control-plane project, or deciding whether sessionstart is useful.
daily-journal
A passive daily work journal that Claude keeps FOR you so you never have to write it yourself. Append short entries after meaningful work (what was done, what you focused on, artifacts touched) to 01-daily/journal/YYYY-MM-DD.md. Run a guided reflection at night or in the morning. Use when you run /daily-journal, say…
doc-drift
Use this skill when the user wants to audit the memory and documents Claude Code loads into context — CLAUDE.md (user global + project + nested), MEMORY.md, @imports, .claude/skills, .claude/agents, .claude/commands, installed plugins — and detect four kinds of issues: outdated claims, mutually contradictory…
ijfw-handoff
Session handoff generation and loading. Trigger: session end, context full, /handoff.