{"id":"21267c27-59ce-4457-a1fa-865e90e6ea7c","slug":"1984-agent","kind":"skill","name":"1984-agent","summary":"Agent skills for team workflows: MR/PR review and testing (standard / hard / ultra)","owner":"tru3ty","repoUrl":"https://github.com/tru3ty/1984-agent","license":"MIT","licenseTier":"permissive","trust":"community","stars":0,"match":null,"fewStars":true,"format":"skill","createdAt":"2026-10-16T12:30:24.000Z","addedAt":"2026-10-22T03:49:57.077Z","logoUrl":"https://avatars.githubusercontent.com/u/125640604?u=c3b9e952ba391b999d0924e125f1cfa75dee7a63&v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["developer-tools"],"tags":[],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-22T20:16:12.181Z","sample":false,"url":"https://conduits.dev/tools/1984-agent","language":"TypeScript","transports":["stdio"],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"# 1984-agent\n\nAgent skills for team workflows. Currently: **mr-review**: a single, team-wide\nstandard for reviewing and testing GitLab merge requests and GitHub pull requests.\n\n## mr-review\n\n```\n/mr-review [MR] [standard|hard|ultra] [--focus <area>]\n```\n\nNo more re-explaining the context for every review. The command picks the MR, the\ndepth and the review areas, runs the project's existing build and tests, and\nprints one report in a fixed format. It posts comments to the MR only after\nyou confirm.\n\n| Mode | How it works | For |\n|---|---|---|\n| `standard` | One agent goes through the whole diff | Small changes that do not touch logic |\n| `hard` | 3 subagents (correctness + intent · architecture · tests + security), merged report | Regular feature MRs |\n| `ultra` | 5 subagents (correctness + intent · architecture · security · performance · tests), then one voting round; the majority sets each finding's severity | Risky or large MRs |\n\nEvery finding states **why** it is a problem, **what it rests on** (a rule,\nan analogous implementation, a spec, a test or a docs link), **why** it has its\nseverity (🔴 blocker · 🟡 should fix · ⚪ nit), and **1–3 ways to fix it** with\ntradeoffs.\n\nImpact sets the colour, proof can only lower it: 🔴 means security, data loss,\na broken spec rule, a broken core flow or a broken build/deploy — and shown by\na reproduction, a failing check or the spec quote next to the breaking line.\nA reproducible cosmetic glitch stays ⚪.\n\nBefore reporting, reviewers check each finding against the code around the\ndiff and the project's own docs (`AGENTS.md`, `CLAUDE.md`, `CONTRIBUTING.md`,\nMR templates, `REVIEW.md` → `Context`), always read from the target branch.\nFindings about a library or runtime API cite the docs of the version the\nproject pins.\n\nThe `intent` role checks that the MR does what its task or spec asks — each\nrequirement quoted next to the code — and nothing it did not ask for.\n\nSmall diffs on risky paths (CI, Dockerfile and compose, deploy scripts,\nmigrations, `auth/`, review rules, `REVIEW.md` → `Risky paths`) get at least\n`hard`. An MR that touches two or more key user journeys or domains (bounded contexts),\nlisted by the team in `REVIEW.md` → `Journeys` / `Domains`, gets `ultra`\nproposed, even when it is small and touches no risky path.\n\n### Guarantees\n\n- **Your working tree is never touched.** The MR code runs in a temporary `git worktree`, which is removed afterwards.\n- **Only failures caused by the MR are blockers.** A test that also fails on the target branch is reported as pre-existing. A failure from a missing DB, secret or network is reported as \"not verified\".\n- **MRs from forks:** the commands are shown before any of the fork's code runs, and nothing runs until you confirm.\n- **Nothing is posted without an explicit answer.** You choose `all` / finding numbers / `no`. Re-runs do not duplicate earlier comments.\n\n### Requirements\n\n- `git`, plus `gh` (GitHub) and/or `glab` (GitLab), logged in.\n- `hard` and `ultra` need an agent with subagents: Claude Code, Codex, Cursor, … `standard` works everywhere.\n\n### Install\n\n**Claude Code (plugin):**\n\n```\n/plugin marketplace add tru3ty/1984-agent\n/plugin install 1984-agent@1984-agent\n```\n\n**Any agent supporting Agent Skills:** copy `skills/mr-review` into the agent's skills folder, e.g.\n`~/.claude/skills/mr-review` (Claude Code) or `~/.agents/skills/mr-review` (Codex and other agents).\n\n### Team rules: `REVIEW.md`\n\nPut a `REVIEW.md` in the repository root to give every review the same context:\nstack, conventions, extra blocker rules, docs to check findings against\n(`Context`), paths that always need a deeper review (`Risky paths`), key user\njourneys and domains that raise the depth (`Journeys`, `Domains`), team and\nlibrary docs (`Docs`), build/test commands, ignored paths and the language of\nMR comments. Start from\n[the template](skills/mr-review/references/review-md-template.md).\n\n## Development\n\n```sh\nbun install\nbun run test                                  # script unit tests + skill structure checks\nsh evals/setup-scenario.sh E2 evals/.work/E2   # behavioural eval fixture, see evals/scenarios.md\n```\n\n## License\n\nMIT\n"}