MySelf is a config collection, not a tool, and the disclaimer says so

One directory per client format, which is the whole design

The repository is a folder per client, and the directory listing makes the structure plain. There are directories named Clash, Egern, Loon, Mihomo, QuantumultX, Shadowrocket, Stash, Surge and sing-box, plus a Script directory, a TikTokUnlock_README.md, an image directory and a .github directory. Each of those client formats has its own configuration syntax, so a rule written for one does not run in another, and the only way to support nine of them from one repository is to keep nine copies. That is the problem the project solves, and it is a real one: a filter list or a rewrite rule that someone else has already written can be read in the dialect of the client you actually use. The repository topics list names the same spread of clients and helpers, from choc, clash and egern through loon, maomi, pharos, quantumultx, shadowrocket, stash and surge to tiktok, which is a good sign that the coverage is deliberate rather than accidental.

The Script directory is the part that carries the real risk, because scripts execute rather than filter, and a rule file can only match on headers and hostnames while a script can rewrite requests, inject certificates or change behaviour in ways a user cannot audit by reading a few lines.

This is a collection of other people's work, credited in full

The README opens with a line that says the content comes from the internet and thanks people for sharing it, followed by a special thanks list of around thirty names and links. Among them are qingr, Semporia, mieqq, app2smile, Maasea, ddgksf2013, xream, zqzess, blackmatrix7, fmz200, Hackl0us, ByteValley, anker1209, Fei, Brian, Coldvvater, Rabbit, Johnshall, Repcz, GetSomeCats, 666OS, Koolson and Orz-3, each linked to a profile, a rules repository, a scripts tree or a set of profiles. That attribution is not decoration, it is the only provenance mechanism the project has. A configuration file for Surge or Shadowrocket that came from somebody else's repository, was adapted, and ended up here has no header telling you who wrote the original rule or under what terms. The credit list in the README is a partial map, not a per-file licence chain, and that distinction matters if you intend to reuse any of it.

The disclaimer is unusually specific, and one line is odd

The disclaimer section is the most substantive prose in the repository, and it is worth reading closely because it defines what you are permitted to expect. It states that the scripts involved are for resource sharing and study and research only, that their legality, accuracy, completeness and effectiveness cannot be guaranteed, and that you should judge for yourself. It says that indirect users, including people who set up a VPS, and anyone whose conduct violates national or regional law or regulation, hold the project not responsible for privacy leaks or other consequences. It asks that nothing be used for commercial or illegal purposes. It says that if anyone believes a script infringes their rights, they should notify the project with proof of identity and ownership, and that the relevant script will be deleted once the documentation is received. It also disclaims responsibility for any script problem, including losses or damage caused by script errors. One clause stands out because it is not a normal licence term: you are told you must completely delete the content from your computer or phone within 24 hours of downloading it. That is a statement of intent about the project, not an enforceable licence, and it sits oddly next to a repository that is still being pushed to.

No releases, no licence, and a push two days ago

The maintenance facts are thin and the release history is empty. The repository has no GitHub releases at all, so there is no version, no changelog and no artefact to pin. The last push to the main branch was on 2026-10-20, so the contents are being changed, but a commit trail without tags tells you nothing about which state of the configuration files was intended to be a release. The licence field for the repository is recorded as unknown, and there is no licence file in the directory listing, unlike most of the projects in this set where a LICENSE file is present. Without a licence, default copyright applies, which means no permission to copy, modify or redistribute is granted. The repository is not archived, so it is not dead, and the frequent pushes mean a configuration you copy today can differ tomorrow. Anyone treating this as a dependency rather than a reading reference is building on a moving target with no stated terms.

Reading a config is safer than running its scripts

The practical difference between the rule directories and the script directory is the difference between a text file you can read and code you have to trust. A filter list is a set of hostnames and rules that the client matches against, and you can read it in a minute to see what it blocks and what it redirects. A script in the Script directory runs in the client environment, and the only way to know what it does is to read its source, which for a redistributed file means finding the original author's version and checking whether the copy in front of you has changed. The disclaimer covers this by stating that it cannot guarantee the accuracy, completeness or effectiveness of the scripts, and by disclaiming responsibility for any script error and the losses or damage that follow. If you want the benefit of a shared rule set without the risk of running unreviewed code, the filter and rewrite configuration is where the value is, and the scripts are where the care belongs.

When writing your own configuration is the better answer

There is a version of this task where collecting other people's configurations is the wrong move. If you use a single client and only need a handful of rules, writing the configuration yourself takes less effort than auditing a file you did not write, and you end up with something you fully understand. The comparison with a per-client maintained repository is instructive. A project that owns one format, such as a rules set written for a single client, can guarantee that the syntax is valid for the version you are running, can be versioned, and can carry a licence. A multi-format collection like this one has to accept that a Surge rule and a sing-box rule expressing the same intent are different files, that one of them will be correct for your client version and the other merely plausible, and that keeping them in step is manual work no automation in the repository performs. The multi-format approach wins on breadth and on discovery. It loses on guarantees, and the empty release history and missing licence are exactly the guarantees you are missing.

Editorial conclusion

MySelf suits an individual user who runs one of the supported clients and wants a starting configuration to read and adjust, and it suits people who want a single place to compare how the same rule set is expressed across clients. It does not suit a team that needs to ship this configuration, because the repository declares no licence, and it does not suit anyone who wants a maintained tool with release notes, since the repository has no releases and the disclaimer places responsibility on the user. Before using anything from it, read the disclaimer in full, check the copyright line of every file you copy, and decide whether the 24 hour deletion clause in that disclaimer affects the way you intend to use the content.

Frequently asked questions

Official sources

1. Issues
2. jnlaoshu/MySelf on GitHub
3. README