# Channel Copy ## Short Launch Post I built Agent Reliability Kit, a local-first scanner for repositories that use AI coding agents. It checks the boring surfaces that often decide whether agents succeed or make a mess: repo instructions, verification commands, README quality, secret hygiene, GitHub Actions safety, MCP/tooling risk, and release readiness. Links: - Repo: - npm: - Docs: If you maintain a repo where Codex, Claude Code, Cursor, Gemini CLI, OpenCode, Copilot, or MCP tools touch code, I would love feedback on confusing findings, missing checks, and install friction. ## X / Threads AI coding agents do better when the repo gives them a reliable operating surface. Agent Reliability Kit scans that surface locally: - agent instructions - build/test/lint commands - README replayability - secret hygiene - GitHub Actions safety - MCP/tooling risk ## LinkedIn I am preparing Agent Reliability Kit, a local-first CLI for maintainers who use AI coding agents in real repositories. The goal is simple: before an agent opens a PR, runs tools, or follows repo instructions, check whether the repository gives it a safe and repeatable path. The scanner looks at agent instruction files, verification commands, README quality, secret hygiene, GitHub Actions safety, MCP/tooling risk, and release readiness. It writes Markdown, JSON, HTML, and SARIF-style outputs so the report can be used locally or in CI. Links: - Public repo: - npm package: - Docs: I am especially interested in feedback from maintainers using Codex, Claude Code, Cursor, Gemini CLI, OpenCode, GitHub Copilot coding workflows, or MCP-based local toolchains. ## GitHub Release / Discussion Draft Agent Reliability Kit is a local-first CLI for checking whether a repository is ready for AI-agent-assisted development. It scans: - agent instructions such as `AGENTS.md`, `CLAUDE.md`, `GEMINI.md`, `CODEX.md`, and Copilot instructions; - repeatable verification commands such as test, build, lint, typecheck, and check scripts; - README quality, including install path, quick start, license, contribution path, and visual proof; - secret hygiene, with token-like evidence redacted in reports; - GitHub Actions safety, including permissions, validation commands, risky triggers, and pipe-to-shell patterns; - AI tooling surfaces such as MCP command configs and prompt-injection-like instruction files. Quick start after public package availability: ```bash npx agent-reliability-kit scan . ``` Until then, the repository can be run from source: ```bash npm install npm run build node dist/cli.js scan . --out .agent-reliability --format markdown,json,html ``` Links: - Public repo: - npm package: - Documentation: Please avoid sharing reports that contain private source paths, private URLs, raw logs, cookies, browser profiles, or real secrets. ## Hacker News / Reddit-Style Post Title: Agent Reliability Kit: local-first repo readiness checks for AI coding agents Body: I built Agent Reliability Kit because many agent failures come from repository surfaces rather than model behavior: missing instructions, unclear verification commands, unsafe CI defaults, README steps nobody has replayed, and accidental secret exposure. The CLI scans a repo locally and reports on agent instructions, commands, README quality, secrets, GitHub Actions safety, MCP/tooling risk, and release readiness. It does not call agent tools or require their credentials. Links: - - - I would value feedback from maintainers using AI agents in real repos, especially on false positives, unclear next actions, and checks that should be added. ## Maintainer Outreach DM Hi. I am preparing Agent Reliability Kit, a local-first CLI that checks whether a repo is ready for AI-agent-assisted development. It scans the surfaces that commonly affect Codex, Claude Code, Cursor, Gemini CLI, OpenCode, Copilot, and MCP workflows: repo instructions, verification commands, README quality, secret hygiene, GitHub Actions safety, and tooling risk. Would you be open to running it on a non-sensitive repo and telling me where the findings are confusing or incomplete? Links: - Repo: - npm: - Docs: ## Product Hunt-Style Tagline Options - Verify and harden AI-agent-assisted codebases in one command. - Local-first readiness checks for repos that use AI coding agents. - Find the repo gaps that make AI agents fail before they open a PR. - A maintainer-friendly scanner for agent instructions, commands, CI, secrets, and docs.