DashClaw
Approval gates and an audit ledger for coding agents left running unattended
Emerged finds new products worth a look, daily — what’s new today
no_human
Open-source agent that takes a ticket to a reviewed pull request, with proof
It treats the coding agent as untrusted: a second model that never saw the coder's session reviews the work, and deleted or weakened tests stop the run before a PR reaches you.
no_human runs on your machine with your own Anthropic, OpenAI or local model credentials. Given a ticket from its board, Jira, Linear or monday.com, it plans, writes the code, runs your tests and opens a pull request; it never merges. An independent reviewer model checks the diff, tamper detection catches removed or weakened tests, and a bug fix must fail on old code and pass on new. When blocked it stops and asks. Over 65 days of building itself it rejected 505 of 1,709 PRs. MIT, v0.2.3, CLI via uv or pipx, desktop apps, a GitHub Action and an MCP server. Paid tiers are unpriced.
More like this
CTRLRun
An open-source safety layer that checks every AI agent action before it runs
TokenEconomizer
Cuts what Claude Code and Codex spend in a repo you work in daily
Sphinx
A pre-push gate that asks what the agent's change does before it ships
Unlazy
An agent skill that refuses to call work done until executable gates pass
OpenBot
Self-hosted AI coworkers, each with its own computer and a policy gateway