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 skills/rasputinkaiser/self-improvement-plugin/sips-selfloopnpx skills add RasputinKaiser/Self-Improvement-Plugin --skill sips-selfloopgit clone --depth 1 https://github.com/RasputinKaiser/Self-Improvement-PluginWhat 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 | $0.00052 | $0.00453 |
| Opus 5 | $0.00026 | $0.00227 |
| Sonnet 5 | $0.00010 | $0.00091 |
| Haiku 4.5 | $0.00005 | $0.00045 |
Grade A, and why
sips-selfloop 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.
What it actually says
SIPS Selfloop
Start or control the persistent loop with python3 scripts/goal_state.py selfloop-set "<focus>" from the SIPS plugin root, following
commands/selfloop.md. Use the homebase_selfloop MCP tool instead when the
live server exposes it.
For each active cycle, establish a measured baseline, select one evidence-backed self-improvement target, checkpoint, make the smallest useful change, verify the gain against the baseline, and record the outcome. Restrict work to SIPS or the agent's reasoning, tools, memory, verification, context efficiency, autonomy, and self-correction. Do not substitute unrelated product work or cosmetic churn.
After a proven gain, run python3 scripts/goal_state.py selfloop-record improved "<proof-bearing summary>" (or homebase_selfloop with action record when
exposed), then record any durable lesson through
SIPS Memory Fabric. Continue immediately while the goal is active. Two
independent plateau cycles may complete the current objective. A real external
block pauses the loop; an explicit stop clears it.
Cost-bounded planning and review
For a planning or research-heavy cycle, use one drafting pass, one independent
blocker audit, and one final validation pass. Give each delegated task a bounded
surface and acceptance check, reuse the same agent for corrections, and do not
start a fresh audit round after every repair. Prefer mailbox completion events to
repeated list_agents/short wait_agent polling.
If the user says to finish, stop, or mentions usage/cost, freeze scope immediately: stop pending expansion, ask active agents for blocker-only results, apply only material fixes, run the minimum gating validation, and hand off. Do not spend another review round polishing an already executable plan.
Report the current cycle, target, baseline, verification, and outcome.
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.
- 2d ago First seen · 40 lines · 52 tokens per session scan A 31aa2fee7228
sips-selfloop is a skill published in the GitHub repository RasputinKaiser/Self-Improvement-Plugin (6 stars, last pushed 5d ago), licensed MIT. It adds 52 tokens to every session and 453 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 2026-08-31.
Other skills, from other repositories
edu-math-tutorial
数学题分步讲解视频的领域知识。适用场景：(1) 用户给出一道数学题并要求做讲解/ 解题视频；(2) 用户说"讲解这道题"、"生成解题视频"、"make a math tutorial"； (3) 需要中文数学教学视频，涉及方程、公式或几何图形。本 skill 只提供领域 知识（题目拆解、讲解节奏、旁白文案规范、版式与 KaTeX 排版风格）；画面、 配音与成片一律用 Creator 原生 Element 与工具从源头构建。.
oma-explanation
Turn a code change (diff, PR, branch, commit range) into a rich, self-contained interactive HTML explainer with Background / Intuition / Code / Quiz sections. Use for explain, walkthrough, code-change explanation, diff/PR/branch explainer requests — 설명서, 해설, コード解説, 代码讲解. Produces a single offline-capable HTML file…
skill-authoring
Author SKILL.md skills: frontmatter, validator limits, structure.
compose:report
Use after implementation is verified and before merge — consolidates multiple spec iterations into a single final-state report, marks related specs, and records key lessons.
skill-creator
Guide for creating effective skills. Use when creating or updating a skill.
openrig-operating-model
Use when you do not know WHERE something belongs: you learned a lesson and are unsure which file takes it; you are about to create a new doc, folder, or convention; you are asking "should this be a skill, a note, or a chain file?"; you inherited a seat and want to know what SHOULD exist; or you are about to duplicate…