falsify-dsh：把公开 Falsify CLI 接进 DSH 做 receipt
`shi275773124/falsify-dsh` 是 DeepSeek Harness (DSH) 的一个适配器，旨在通过对接公开的 Falsify CLI 来弥补 Agent 自我判断与可核对发布标准之间的缺口。该插件不生成第二意见，而是将 CLI 的 `PASS`、`PASS_WITH_DEBT` 或 `BLOCK` 等回执（receipt）直接交还，确保判决由 CLI 做出而非 Agent 自行推断。 核心功能包括三个工具：`falsify_lint` 用于静态检查，返回清洁度标签；`falsify_review` 用于文档审查，返回最终判决；`falsify_gate` 用于 PR
Read Moredsh-simplify: A DSH Plugin for Reviewing and Simplifying Code Within the Scope of Changes
dsh-simplify is an MIT-licensed plugin for DeepSeek Harness, maintained by GongYuanCaiJi, designed to address the need to clean up only the modified portions of code after changes, rather than rewriting unrelated files. Ported from pi-simplify, the plugin parses the git diff to precisely identify changed files and line number ranges, restricting the Agent to performing simplification, clarity, and maintainability optimizations only within the changed lines. Its core features include strictly limiting the editing scope to the changed lines, supporting four review modes—uncommitted changes, staged changes, comparison against a specified ref, and specified files—and running existing tests before the review concludes to verify that modifications are correct. It also automatically adheres to the project's CLAUDE.md or AGENTS.md conventions. Installation requires Node.js 20.
Read Moredsh-revdiff: DeepSeek Harness Interactive Diff Review Plugin
`dsh-revdiff` is a native interactive Git diff review plugin for DeepSeek Harness, designed to enable a closed-loop workflow of "modifying code - reviewing changes - providing feedback to the Agent". The plugin is released under the MIT license and requires a Node.js environment of ^22.19.0 or >=24.0.0. Its core functionality includes supporting interactive reviews of tracked, staged, or unstaged changes, allowing annotations such as issues, suggestions, questions, and praise to be added at the file and code-line level, and submitting structured review comments directly to the current Agent via `agent.followup()`. Users can invoke it with the `/revdiff` command and parameters such as `--staged`, paths
Read Moredsh-commit-review: Adding /commit and /review to the DSH Web GUI
dsh-commit-review is a DSH plugin maintained by the-qian, designed to enhance the git commit and review workflow through a Web GUI. The plugin adopts a pure Host-side zero-dependency architecture, requiring no build steps, and integrates via the `dsh.bundle.patch` format. It primarily provides two slash commands: `/commit` analyzes current git changes and actually creates commits, supporting the Conventional Commits standard and custom prompts, while adhering to safety conventions such as not modifying config, not forcing, and not skipping hooks, and strictly prohibiting the commit of secret files; `/review` reviews uncommitted changes, outputting Markdown reports graded by severity, with support for security, performance, style, and other aspects.
Read Moredsh-skill-7d-code-reviewer: Adding Templated Code Review to DSH Sessions
`7dgroup-ai/dsh-skill-7d-code-reviewer` is a template-driven code review plugin designed specifically for DeepSeek Harness (DSH). It aims to address issues such as unstable review scope, inconsistent levels, and format inconsistencies caused by ad-hoc prompts. Licensed under MIT, the current version is `0.1.0-rc.5`, and it requires Node 22+ and pnpm 10+ environments. Installation follows GitHub specifications to circumvent build script restrictions. Its core capabilities include: separating review content (SKILL.md) from presentation templates (HTML); executing a five-step process that involves task acceptance, quick scanning, line-by-line review, grading, and report generation; and providing fixes based on three severity levels—Critical, Medium, and Minor.
Read Moredsh-file-review: Review Agent-Modified Files in DSH Web
dsh-file-review is a DeepSeek Harness Web workflow plugin maintained by left0ver, which allows users to review files modified by an agent within the browser via a Diff panel. It supports Standard and PTC modes, line-level comments, partial rollbacks, and both Chinese and English interfaces. This article introduces its positioning, core features, pnpm whitelist configuration for cooldown periods, as well as installation commands such as "dsh plugin --profile web add" and typical usage workflows like Chat→Generate→Review. It also reminds users that the plugin runs with the current dsh process permissions and advises verifying the source code and MIT license before installation.
Read More# workers-best-practices Skill: Review and Write Cloudflare Workers in Accordance with Production Agreements
Cloudflare's official workers-best-practices Skill is used to review and write Workers code in accordance with production conventions: it pulls current documentation and types first, then checks `wrangler.jsonc`, streaming responses, floating promises, global state, secrets and observability. This article verifies the installation method, rule list and anti-patterns based on the original GitHub content and the official best practice page, and explains that Durable Objects, Workflows and the wrangler CLI need to switch to other Skills in the same repository.
Read More`find-bugs: Let AI scan the current branch for bugs and security issues according to the checklist`
find-bugs is an Agent Skill publicly released by the Sentry Engineering Team in the getsentry/skills repository, which is used to review the changes made in a local branch relative to the default branch. It executes in five stages: obtaining the complete diff via Git and GitHub CLI, mapping the attack surface, going through 11 security checklists such as Injection, XSS, authentication, and IDOR item by item, verifying false positives, and finally generating a report with evidence, while explicitly stating that it will not modify any code. It can be installed via the Claude Plugin Marketplace or `npx skills add`, and is compatible with tools that support SKILL.md such as Cursor, Claude Code, and Codex.
Read MoreHN Hot Topic: Can Manually Re-typing Every Line of Code Generated by an LLM Avoid "Cognitive Debt"?
On August 4, 2026, Ankur Sethi published a post advocating for manually retyping LLM-generated code line by line to avoid "cognitive debt", which garnered 409 points and 348 comments on Hacker News. This article outlines its core workflow (the Agent only displays changes without directly writing files), the pros and cons debates on HN, as well as more widely accepted alternatives like plan-first, design-first, and test-driven development, to help developers balance efficiency and code comprehension in the era of AI programming.
Read MoreGitHub Native Support for Stacked PRs: Splitting AI-generated Giant PRs into Audit-friendly Small-step Commit Chains
On July 30, 2026, GitHub announced that Stacked Pull Requests had entered Public Preview. On August 4, the engineering blog detailed how to use the gh stack CLI and Agent Skill in conjunction to split large PRs generated by AI in one go into multiple small layered PRs according to dependencies. This article sorts out the pain points of reviewing giant PRs, the layered thinking, CLI getting-started commands, the Stack Map review strategy, as well as precautions for rebase and signed commits.
Read MoreGitHub Stacked PR Public Beta: Split large changes into independently reviewable small layers, merge the entire stack with one click
On July 30, 2026, GitHub announced that Stacked Pull Requests had entered public beta. This feature allows large code changes to be split into an ordered chain of PRs, supports parallel code reviews and one-click merging of the entire stack. Based on the official changelog and documentation, this article introduces the dependency chain structure of stacked PRs, the getting-started steps of the gh stack CLI, the web-side stack map, the integration of Copilot gh-stack skill, as well as Merge Queue support and usage restrictions.
Read MoreGitHub Stacked PR Public Beta: Split large changes into PR chains that can be independently reviewed and merged with one click
On July 30, 2026, GitHub announced that Stacked Pull Requests had entered public beta, supporting the splitting of large changes into an ordered PR chain, independent layer-by-layer reviews, and one-click merging of the entire stack. Based on official changelogs and documentation, this article introduces the branch dependency model, review and CI mechanisms, the getting-started steps for the gh-stack CLI, as well as the collaboration methods with Merge Queue and Copilot Agent and current limitations.
Read MoreGitHub Stack PR Official Public Beta: gh stack CLI lets you split large changes into chains of independently reviewable small PRs
On July 30, 2026, GitHub announced that Stacked Pull Requests entered public beta. The supporting gh stack CLI extension allows splitting large changes into a chain of dependent small PRs, enabling independent review for each layer and one-click merging of the entire stack, with native integration with Branch Protection and Merge Queue. Based on the official Changelog and documentation, this article introduces the concept of stacked PRs, installation commands, the workflow from init to submit, as well as the collaboration method with Trunk-based Development.
Read MoreCopilot Code Review Integrates Agent Skills and MCP: Team Guidelines Finally Can Be Included in Reviews
On July 29, 2026, GitHub announced the general availability of Copilot code review support for Agent Skills and MCP servers, which is now available to Pro, Business and Enterprise users. Teams can inject internal coding and review specifications via the SKILL.md file in the .github/skills directory; MCP can pull external contexts such as Jira and document systems in read-only mode, and attribution tags for skills and MCP are added to review comments. This article sorts out the core capabilities of the general release, the configuration examples of SKILL.md, the read-only security boundary of MCP and implementation suggestions.
Read More