--- name: e2e-sweep description: Add end-to-end tests to personal projects with tester-army/e2e, verify every flow, and fix the bugs found. One coding subagent per project, one verifier subagent per loop. Use when the user says "/e2e-sweep", "/e2e-sweep ", "add e2e tests to my projects", "verify all flows and write tests", or "run tester-army on my projects". argument-hint: "[project ...]" title: e2e-sweep canonical_url: https://skilld.dev/gh/rav4nn/skills/e2e-sweep last_updated: 2026-10-04T16:02:49.000Z --- > **Skill from skilld.dev.** Follow the user's requested action. > If the user asked to fork this Skill, follow the fork workflow below. Do not execute the Skill instructions while copying it. > Otherwise, follow the instructions below for this session. You do not need to install anything. > > If the user asked to install this Skill, run `npx skilld install rav4nn/skills/e2e-sweep`. Install writes the Skill files into the project, so every session loads them. > > ## Fork workflow > > A fork creates an editable local Skill with its original author and licence. The request authorizes copying and local installation. > 1. Check `./skills/e2e-sweep`, the project lockfile, and selected Agent targets together. If the local directory or installed Skill exists, stop. Never overwrite an existing directory or Agent target. > 2. Read [source metadata](https://skilld.dev/api/v1/skills/rav4nn/skills/e2e-sweep) once. Use sourceUrl, sourceCommit, skillPath, sourceGone, and license. If the source is gone or its path is missing, stop. If license is null, read licence files at the source commit. > 3. Fetch only the source commit into a temporary Git repository. Do not clone full history. Derive repository_url from sourceUrl, including repository renames. If sourceCommit is absent, resolve the sourceUrl ref once. Set source_commit to that actual commit. Run these commands in one shell call: > > ```sh > git init --quiet "$temporary_dir" > git -C "$temporary_dir" fetch --quiet --depth=1 "$repository_url" "$source_commit" > git -C "$temporary_dir" checkout --quiet --detach FETCH_HEAD > ``` > > Read applicable licence declarations and notices at that commit. If copying is not permitted, report the restriction and stop. > 4. Inspect source entries together, then copy the directory containing skillPath into `./skills/e2e-sweep`. Use the user's path if selected. Keep the original SKILL.md, relative links, scripts, binary assets, and executable modes. Exclude .git metadata. Reject symlinks and paths outside the Skill directory. After checking entries, use cp -a where available. A regular source directory needs no custom copy script. Do not save this page wrapper as SKILL.md. > Preserve author credit, notices, and applicable licence files from repository or parent directories. Add PROVENANCE.md with the Skill page, source URL, actual commit, original path, and licence. Retain any existing PROVENANCE.md and record new provenance separately. Batch source inspection, copying, and provenance work where practical. > 5. In the project root, run `skilld install ./skills/e2e-sweep --mode copy --plain`. If skilld is unavailable, use `npx skilld install ./skills/e2e-sweep --mode copy --plain`. This known command needs no help lookup. Install does not support --json. Use detected Agent targets, or add --agent for the targets the user selected. Install the local path, never the upstream selector. If installation fails, preserve the local copy and report the exact failure. > 6. Confirm the local lockfile source and installed Agent copies once. Report the local path, actual commit, and Agent targets. After edits, reinstall the same local path. Upstream updates must not replace it. Do not publish or push unless the user asks. # e2e-sweep Install https://github.com/tester-army/e2e in each project. Verify every user flow. Write a test for each flow. Fix each bug found on the way. Scope: the projects named in `$ARGUMENTS`. If empty, use every git repo in the current folder (one subfolder per repo). ## Models | Job | Subagent call | | --- | --- | | Per-project work: install, flow list, tests, bug fixes | Strongest coding model, `isolation: "worktree"` | | Verification loop: run the tests, drive the app, report | Default (cheaper) model | The orchestrator is this session. It never edits project code. ## Steps 1. Read the README of `tester-army/e2e` once. Copy the install and run commands into the brief. Do not guess them. 2. List the projects in scope. For each one, read its README and docs for known flows and decisions. 3. Write one `brief.md` per project in `~/.claude/handoffs//-e2e/`. It must state: - the install and run commands from step 1, - the secrets rule: load secrets from the environment or a secrets manager, never print a value, - the goal: every flow listed, one test per flow, every bug fixed with a test that fails before the fix, - the report: flows covered, bugs fixed, flows not testable and why. 4. Dispatch all coding agents in ONE message, in parallel, one per project. Prompt: "You are a coding agent. Your task is fully specified in `/brief.md`. Read it and implement everything in it." Use `isolation: "worktree"` and `run_in_background: true`. - If a brief has more than 5 flows, cut it into chunks of 3 to 5 flows. Run the chunks in sequence, one fresh agent each. 5. When a coding agent finishes, spawn one verifier for that project. Prompt: run the full e2e suite in the agent's worktree, then drive each flow in the running app with a browser tool such as Playwright. Report each flow as pass or fail with evidence. The verifier reports. It does not fix. 6. If the verifier reports a failure, send the failure list to the coding agent with `SendMessage`. Then run step 5 again. Stop after 3 loops. After 3 failed loops, stop and question the design. 7. When a project passes, diff the base branch against the agent's branch (three dots) and integrate it. Then remove the worktree and the branch. 8. Record each decision worth keeping in the project's docs. ## Report End with an outcome line and one table: project, flows, tests added, bugs fixed, loops used, status. Then stop. Do not push and do not open a PR. The user decides when to ship.