Simulates persona-driven user journeys through a live product UI using Playwright tests.
npx skills add https://github.com/letuhao/lore-weave --skill human-sim-letuhao
Please help me install this Agent Skill. Skill: human-sim Source: https://github.com/letuhao/lore-weave/tree/main/.claude/skills/human-sim Command: npx skills add https://github.com/letuhao/lore-weave --skill human-sim-letuhao
newcomer on a fresh empty account and a frequent user seeded to at least 25 books — catch empty-state problems and scale-related defects like pagination and truncation bugs. - Evidence tooling: Per-step x-trace-id labels, a log collector (collect_run_evidence.py) that exits non-zero when a correlated query matches nothing, and a humanRun.ts helper that captures a full-page snapshot per step, pass or fail. - Calibration discipline: Every journey must be validated by deliberately breaking the behavior it protects and watching the test go red, ensuring checks can actually fail. - Use Case: A user with 83 books searches their library by exact title and gets nothing because the frontend filters only the first 20 loaded rows; the frequent persona journey catches this, while a fresh-account suite would report success forever. ## Quick Start Run the persona journey Playwright spec against the local lw-iso frontend on port 25174 and review the per-step snapshots and collected logs for any failing claim.High-intent search queries and answers about installing and using this skill.
Run npx playwright test tests/e2e/specs/persona-journeys.spec.ts from the frontend directory with PLAYWRIGHT_BASE_URL set to http://localhost:25174. The journeys write data, so the target must be loopback; non-loopback targets are refused.
Scale-dependent defects like search filtering only the first loaded page do not exist below 21 books. The minBooks requirement is enforced by seeding, so the journey actually exercises pagination and truncation behavior a small account cannot reach.
The documented account credentials belong to the infra stack on port 5174, not the lw-iso stack on 25174. This is a wrong-stack error, not a broken login; use the correct stack's credentials.
Provider credentials and default models are configured per owner_user_id, so a new account has no model and the assistant cannot write prose. Use the documented account with a throwaway book for generation journeys.
A correlated log query that matches nothing exits 2 rather than 0, because an empty result with a green exit would falsely suggest a clean step. Check that the frontend image predates the trace header, the id was pinned, and the id was not malformed.
Rebuild before testing your own change, because the container serves a baked build and an un-rebuilt image silently tests the previous commit. Verify by grepping the running container for a string only your change contains, not by trusting the build log.
Pythagora-ioProvision isolated local development environments with per-feature API, frontend, and database.
FlintmereReproduce reported defects in Flintmere's codebase and validate fixes with regression tests.
CODE-SAURABHGenerate automated tests for unit, integration, end-to-end, contract, and snapshot testing.
woinkAudit codebases and generate a .harness/CODEBASE_UDIT.md planner-ready assessment with Context7 dependency verification.