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.
git clone --depth 1 https://github.com/madebyaris/advance-minimax-m3-cursor-rulesWrote 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/rules/madebyaris/advance-minimax-m3-cursor-rules/minimax-m3-self-evolution)<a href="https://agentmods.dev/rules/madebyaris/advance-minimax-m3-cursor-rules/minimax-m3-self-evolution"><img src="https://agentmods.dev/badge/rules/madebyaris/advance-minimax-m3-cursor-rules/minimax-m3-self-evolution/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/rules/madebyaris/advance-minimax-m3-cursor-rules/minimax-m3-self-evolution"><img src="https://agentmods.dev/badge/rules/madebyaris/advance-minimax-m3-cursor-rules/minimax-m3-self-evolution.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 | $0.00026 | $0.00760 |
| Opus 5 | $0.00013 | $0.00380 |
| Sonnet 5 | $0.00005 | $0.00152 |
| Haiku 4.5 | $0.00003 | $0.00076 |
Grade A, and why
minimax-m3-self-evolution 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 — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MiniMax M3 Self-Evolution Harness
M3 has a 1M-token MSA context and high skill adherence, and supports autonomous self-improvement through iterative reinforcement learning loops. Use this rule when the task involves debugging, optimization, or recursive code improvement.
Core Loop
Iterate until evidence shows the problem is solved or the approach is exhausted:
1. Run the smallest diagnostic check
2. Read the failure output directly
3. Compress prior raw evidence that you no longer need (see below)
4. Make ONE targeted fix based on evidence
5. Re-run the exact check that failed
6. If fixed, verify the broader surface
7. If not fixed, form a new hypothesis from the NEW evidence only
Do NOT repeat the same fix twice on the same hypothesis. Do NOT assume the cause without reading the evidence.
Compress Before Next Iteration (M3)
On M3 the failure mode shifts from "ran out of room" to "ran too many parallel hypotheses without compressing." Before each new iteration:
- Replace raw search/fetch output from previous iterations with a 2–4 line summary.
- Drop evidence that is no longer relevant to the current hypothesis.
- Keep one canonical "current best hypothesis" line at the top of your scratchpad.
- For very large work, use the
minimax-m3-long-contextskill to plan the loader.
When to Use This Harness
| Task | Harness Approach |
|---|---|
| Debug runtime error | Read error → one fix → re-run |
| Optimize performance | Profile → smallest change → measure |
| Fix failing test | Read test → read code → one fix → run test |
| Investigate unexpected behavior | Log/check → hypothesis → targeted probe |
| Iterative code improvement | Current state → one improvement → verify |
| Triage a visual bug from a screenshot | Read the screenshot, form hypothesis, one fix, re-read the post-change frame (multimodal-grounded) |
Iteration Limits
- Diagnostic phase: 3 iterations to identify root cause
- Fix phase: 2 iterations per hypothesis before switching strategy
- Overall: If 5 total iterations pass without progress, summarize evidence and ask
- Context check: if you have run 3+ raw search/fetch operations without compressing, compress before continuing
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 · 82 lines · 26 tokens per session scan A 7c6307834111
minimax-m3-self-evolution is a cursor rule published in the GitHub repository madebyaris/advance-minimax-m3-cursor-rules (125 stars, last pushed 2mo ago), licensed MIT. It adds 26 tokens to every session and 760 once invoked, about $0.0001 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.
Other cursor rules, from other repositories
debugging
Use this when debugging failures, broken tests, runtime errors, or unexpected behavior.
performance
Performance: profiling-first optimization, caching, bottleneck patterns.
error-handling
Error handling: typed errors, boundaries, user messages.
git-commit
Use this when the user asks for a git commit, commit message, or commit command.
learnship
Activate learnship agentic engineering workflows — structured phases, persistent memory, learning partner, and impeccable UI design. Use whenever a user is working on a software project or asks about project planning, phases, workflows, or how to build something.
coding
Core coding behavior for this repository. Always prefer cautious, minimal, reversible changes.