reckless: rust rewrite by daywalker90 · Pull Request #9239 · ElementsProject/lightning

daywalker90 · 2026-06-19T16:48:24Z

Not ready for code review yet, we have not decided on some important aspects. Right now it is a drop-in replacement of the python reckless version.

Should
Fixes: #8439
Fixes: #8761
Fixes: #8775
Fixes: #8776
Fixes: #8794
and should supercede #8630

Key differences:

it is now a plugin only, the only use case to have a standalone binary, that i could come up with, is a broken plugin install where CLN won't start, but if you can call a binary you can uncomment a line in a config, right?
as a consequence, alot of options are no longer necessary and reckless-dir is a proper plugin option now
except for the help we always output json, so no more --json option
the rewrite does not adhere to the reckless schema in all places, e.g. listavailable, an array of strings is just not nice here
a tip function was added to tip plugin authors that have an offer in their manifest.json

vincenzopalazzo

Nice to see that the manifest idea is taking over. Hope you are also using git for versioning, otherwise this would be another good thing to take from https://github.com/coffee-tools/coffee

vincenzopalazzo · 2026-06-19T18:46:40Z

Would be nice to reference the original reference implementation https://coffee-docs.netlify.app/support-coffee

Would be nice to reference the original reference implementation https://coffee-docs.netlify.app/support-coffee

I was planning on giving full credit for the manifest/tip function to coffee in the README. But i want to lock down the functionality first before writing documentation.

to save credit was also good to just include coffee, I do not think there are more features than coffee, we were also supporting ws (palanning to) to support interactive UX inside the web applications

But these days cloning is easy!

daywalker90 · 2026-06-19T18:50:18Z

Nice to see that the manifest idea is taking over. Hope you are also using git for versioning, otherwise this would be another good thing to take from https://github.com/coffee-tools/coffee

If you mean installing a plugin with a git ref then yes, that is supported.

vincenzopalazzo · 2026-06-19T18:53:28Z

If you mean installing a plugin with a git ref then yes, that is supported.

Yeah, otherwise it's insanity! Nice, it would be nice to have two lines that reference coffee because the development died because Blockstream was moving to reckless. Referencing a good idea is what is left in this AI era, where rebuilding is easy

daywalker90 · 2026-07-14T13:24:22Z

~~Some tests will fail for now because of #9283. If we decide not to fix it, i can do a workaround.~~

Changelog-Changed: reckless: complete rewrite with new subcommands `listinstalled` and `tip`. Replaced `search` with `listavailable`.

ScuttoZ · 2026-07-23T14:47:26Z

+            return Err(anyhow!("install_cmd in manifest is empty!?"));
+        }
+
+        let mut cmd = Command::new(cmd_parts.first().unwrap());


install_cmd in the reckless manifest lets any plugin repo specify a list of commands that reckless install will run on the user's node, with no confirmation prompt and no sandbox, contrary to the previous installation behaviour which only ran commands from a list of authorized installers (cargo, uv, poetry, npm, etc..).
Should we treat this kind of user-added sources differently? Would it be wise to keep this but adding a flag that explicitly allow this?

Not sure a flag would improve security or anything. Unless something else on your machine is worth more than your node's Bitcoin and these commands could get access to that. Otherwise you already have to trust the plugin author's code to not do something nefarious like calling xpay with an offer to the plugin author's node until your balance is drained.

But isn't that a big change in code capabilities if the installation process was previously sandboxed and now is not? I'm not referring to an improper use of the node, but to potential arbitrary code execution. Is this prevented in some way I'm currently missing?

Not sure what you mean by "change in code capabilities"? Was the python version doing something like that?

The install commands are read from the manifest.json file included in the plugin's source code. An attacker can write an attack in there or in the plugin itself. Both will run commands under the CLN user, i see no significant difference between the two.

oh so this is not a new attack vettor, my bad. I must have misinterpreted how install_to_python_virtual_environment worked in the old reckless python file. I thought the provided code was somehow vetted/sandboxed. Nevermind then, I withdraw my first comment :)

ScuttoZ · 2026-07-24T12:59:26Z

+        }
+    }
+
+    match fs::remove_dir_all(&plugin_dir).await {


uninstall reclaims reckless_dir/<plugin_name> but never the cloned source repo at reckless_dir/<owner>/<repo>. Is this intended, to work as some sort of cache?

Yes, this is intended. Checking out a repo can take some time so ideally we only want to do it once when the source is added. So then we also only remove it when the source is removed not a plugin belonging to that source. We would also have to check if the repo contains a single plugin or multiple for this to make sense.

If the timestamp of the python file does not differ after an update it will load the bytecode of the old version. We remove the pycache to make sure this does not happen. Mostly a test environment issue as in practive the timestamps should always differ.

…rning an error

daywalker90 force-pushed the reckless-rs branch 2 times, most recently from 8644166 to 6889d36 Compare June 19, 2026 17:21

vincenzopalazzo reviewed Jun 19, 2026

View reviewed changes

madelinevibes added this to the v26.09 milestone Jun 22, 2026

madelinevibes assigned daywalker90 Jun 22, 2026

daywalker90 force-pushed the reckless-rs branch 3 times, most recently from 6658af8 to f6bbfb5 Compare July 1, 2026 17:09

daywalker90 force-pushed the reckless-rs branch 5 times, most recently from 5833d8b to 853fe7d Compare July 8, 2026 15:34

daywalker90 force-pushed the reckless-rs branch 4 times, most recently from f4513c8 to bf9e3e2 Compare July 14, 2026 12:39

daywalker90 marked this pull request as ready for review July 14, 2026 13:11

daywalker90 requested a review from cdecker as a code owner July 14, 2026 13:11

daywalker90 force-pushed the reckless-rs branch from bf9e3e2 to 2dde507 Compare July 15, 2026 14:59

madelinevibes added the QA Blockstream QA team have reproduced, or a test has been created! Look for the linked PR/Issue label Jul 17, 2026

daywalker90 force-pushed the reckless-rs branch from 2dde507 to 59f0030 Compare July 17, 2026 12:05

reckless: rust rewrite

a0ca360

Changelog-Changed: reckless: complete rewrite with new subcommands `listinstalled` and `tip`. Replaced `search` with `listavailable`.

daywalker90 force-pushed the reckless-rs branch from 59f0030 to a0ca360 Compare July 21, 2026 11:04

madelinevibes unassigned daywalker90 Jul 23, 2026

madelinevibes added the Status::Ready for Review The work has been completed and is now awaiting evaluation or approval. label Jul 23, 2026

madelinevibes requested a review from enaples July 23, 2026 07:59

madelinevibes requested a review from ShahanaFarooqui July 23, 2026 07:59

ScuttoZ reviewed Jul 23, 2026

View reviewed changes

ScuttoZ and others added 3 commits July 24, 2026 11:05

typo: changed "lintintalled" instances to "listinstalled"

2b0c4c4

fixup! typo: changed "lintintalled" instances to "listinstalled"

9d8ce25

fix repo owner dir removal when removing source

e71308e

ScuttoZ reviewed Jul 24, 2026

View reviewed changes

daywalker90 added 5 commits July 29, 2026 12:19

Makefile: add cln-rpc and cln-plugin as dependency of reckless

883f4b9

tests: add reckless test for update

fe3a970

tests: add reckless tests for listinstalled

bf6d5bb

reckless: listinstalled: check if reckless dir exists instead of retu…

a1ce5dd

…rning an error

daywalker90 mentioned this pull request Jul 29, 2026

reckless returned invalid character in json output daywalker90/summars#55

Closed

Uh oh!

Conversation

daywalker90 commented Jun 19, 2026

Uh oh!

vincenzopalazzo left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

daywalker90 commented Jun 19, 2026

Uh oh!

vincenzopalazzo commented Jun 19, 2026

Uh oh!

daywalker90 commented Jul 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

daywalker90 commented Jul 14, 2026 •

edited

Loading