Taskless vs Writing the Hook Yourself
The usual way to make a coding agent follow a rule is a husky or pre-commit hook running a lint rule you asked a model to write. If you already know the rule, do that. Taskless is for finding the rules you have not written yet, and for proving that each one catches what you meant.
Open source CLI. Runs locally, no account required.
Writing the rule was the easy part
A developer, a frontier model, and an afternoon will get you a lint rule. Wire it into a husky or pre-commit hook and the agent cannot commit past it. For a rule you already know, that is a good answer, and it is the one Taskless gives too: when your linter can express a rule, the rule goes into your linter.
Two things are harder than writing the rule.
Knowing which rules to write
The rules that matter are the corrections your team keeps making in review: the client everyone is told to use, the error that must be logged, the field a migration always forgets. They are spread across thousands of review comments, and nobody has time to read through them. So most never become rules, and they stay as comments a reviewer has to repeat.
Taskless reads your review comments, fixup commits, and agent files, finds the corrections that repeat, and writes those as rules.
Proving the rule is right
A rule that flags the wrong code gets ignored, and a rule that misses the case you meant gives false confidence. Asserting correctness is the hard part. Every Taskless rule ships with passing and failing fixtures, and the CLI runs them before the rule is recorded. When the rule needs to change, the fixture for the case that was wrong comes with the fix.
At a glance
| Taskless | Hook plus a hand-written rule | |
|---|---|---|
| Which rules to write | Found in your review comments, fixup commits, and agent files | The ones you remember to write |
| Proof the rule is right | Passing and failing fixtures, run before the rule is recorded | Whatever tests you write for it |
| Where the rule lives | Your linter when it can express it; otherwise ast-grep, Vale, or runtime | Wherever you put it |
| How it runs | taskless check in your hook, in CI, and in the agent’s session |
Your hook |
| Getting started | Open source CLI, no account | An afternoon per rule |
Use both
This is not a choice between Taskless and hooks. Hooks are where checks belong, and taskless check runs in one like any other linter. Write the rules you already know by hand if you like. Let Taskless find the rest and prove them.
Claims about DIY pre-commit hooks last checked .
Frequently Asked Questions
Why not ask Claude to write me an ESLint rule?
Does Taskless replace my pre-commit hook?
How does Taskless find rules I have not written?
Write the rule once. Check it every run.
Install the CLI, then ask your agent to onboard the project. It reads your codebase and history and writes the rules it can enforce locally.