Devotd Github
NewCoding agents

devanshranjan10/landmines v0.2.0 — a tool that warns agents about dependency bugs

devanshranjan10/landmines is a TypeScript tool that links dependency versions in the lockfile with public GitHub issues, feeding known problems and workarounds into an agent's workflow.

Repositorydevanshranjan10/landmines ↗

devanshranjan10/landmines is a TypeScript tool that links dependency versions in the lockfile with public GitHub issues, feeding known problems and workarounds into an agent's workflow.

GitHub description: Known open bugs in your exact dependency versions, put where your coding agent reads them (AGENTS.md/CLAUDE.md, pre-edit hook for 20 agents, MCP).

stars
★ 7
Forks
0
Language
TypeScript
License
MIT
Latest release
v0.2.0
Read original source ↗View on GitHub ↗

devanshranjan10/landmines is a TypeScript tool that connects the dependency versions pinned in the lockfile with public GitHub issues to find known problems. The latest release is v0.2.0. The tool alerts agents to the risks it finds before they edit files, and also provides ways to avoid the problems. However, the provided record contains no specific release notes about what changed in v0.2.0.

The problem information can be recorded in CLAUDE.md or AGENTS.md, which agents read, and can also be delivered through a pre-edit hook and MCP. The key point is that it links the exact versions written in the current repository's lockfile to the issues, rather than giving generic bug descriptions. The repository description states that the pre-edit hook targets 20 agents. According to the project's report, even with web search or a docs MCP, Sonnet 5.5 failed to find known bugs, but after being given the bug list it passed 83 of 84 questions.

This result is an evaluation reported by the project, and does not mean the same effect appears in other repositories or environments. How completely it links dependency versions and issues, and whether the warnings actually change agent behavior, must be verified separately. For now, it is reasonable to regard it as an aid that surfaces known risks before an agent starts working.