web-scenario-tester

Plugin web-scenario-tester in Emasoft/ai-maestro-web-scenario-tester. Commands read from the repo's manifest — 1 month ago.

Skills (14)

• amwst-create-scenario

  Use when adding a new scenario. Trigger with "create a new scenario", "write a scenario for X", or "author SCEN-N". Assigns the next number, drafts frontmatter/phases/steps, saves the file.

• amwst-edit-scenario

  Use when modifying an existing scenario file. Trigger with "edit SCEN-N" or "add a step to scenario N". Adds/removes/reorders steps, tightens verifications, fixes Rule 6, bumps version.

• amwst-implement-scenarios-proposals

  Use when proposal files need P0 items applied to code. Trigger with "implement proposals from scenario N" or "fix P0 issues from last batch". Spawns implementer in an isolated git worktree.

• amwst-improve-scenario

  Use when a scenario needs post-run analysis. Trigger with "improve SCEN-N", "analyze SCEN-N reports", or "strengthen flaky steps in N". Identifies patterns and proposes changes.

• amwst-phase-execute

  The EXECUTE phase of a UI scenario run — how to drive the steps cheaply. The amwst-scenario-runner loads this when it reaches step execution (after SAFE-SETUP). Covers fixed-first load order, reading ONE step at a time with amwst-scenario-step.sh (never re-reading the whole .scen.md), scoped snapsho

• amwst-phase-fixasyougo

  The FIX-AS-YOU-GO phase of a UI scenario run — when a step fails because the app under test has a real bug, diagnose from real data, fix the root cause, rebuild, and retry the same step, cheaply. The amwst-scenario-runner loads this when a step assertion fails. Covers scoped source reading, the amws

• amwst-phase-proposals

  The 11th-HOUR PROPOSALS phase of a scenario run — produce concrete, prioritized improvement proposals AFTER a scenario has run. Loaded by the amwst-scenario-proposer agent (a SEPARATE agent from the runner that did fix-as-you-go). Reads the run's structured report + the scenario greppably, analyses

• amwst-region-capture

  Token-cheap visual verification for UI scenarios. Capture DOM/ARIA-scoped CLIPPED screenshots (target element + margin) and SCOPED aria snapshots instead of full-page captures — using the live ARIA/DOM tree as the guide for what to clip. Use whenever a scenario step must visually check CSS/layout (t

• amwst-run-scenario

  Run ONE (or several) UI scenario test end-to-end in an isolated forked subagent context. Use when the user says "run scenario N", "execute SCEN-018", "run the maintainer scenario", "test scenario 1", "rerun 1 and 19", "run the title-change scenario", or anything that names a SCEN file. Forks one amw

• amwst-run-scenarios-batch

  Use when the user wants unattended batch execution of scenario files. Trigger with "run scenarios 16-20" or "batch 5-10 --improve". Accepts a range, comma list, or single number.

• amwst-scenarios-rules

  The 14 mandatory scenario rules every runner MUST follow. Use when running, editing, or implementing a scenario. Trigger with "run scenario NNN" or "execute SCEN-NNN", or preloaded via subagent skills frontmatter.

• amwst-step-batch

  Run MANY scenario UI steps in ONE dev-browser call (one turn) instead of one tool-call per action. A declarative {action, assertion} step list executes sequentially in a single sandbox execution and STOPS at the first failed assertion, so FIX-AS-YOU-GO stays intact. Use to collapse the snapshot->act

• amwst-validate-scenario

  Validate a .scen.md scenario file with amwst-validate-scenario.py before running or committing it. Use after writing or editing a scenario (amwst-create-scenario / amwst-edit-scenario), or when the user says "validate the scenario", "lint SCEN-012", "check my scenario file is well-formed". Reports e

• the-skills-menu

  Dynamic skill menu for the web-scenario-tester plugin. Teaches agents which skills are available, when to use them, and how to load them with the Skill() tool. Use when an agent needs to pick a downstream skill at runtime. Used by every web-scenario-tester agent via the-skills-menu method (TRDD-9dd6

Subagents (6)

• amwst-parallel-tester-agent

  Runs a focused smoke-test (≤10 UI steps) against the currently-running web app under test via dev-browser to verify a feature the amwst-parallel-worker-agent just merged. Returns a 2-line pass/fail summary so the orchestrator can decide to resume the long scenario run or spawn a fix cycle. Unlike am

• amwst-parallel-worker-agent

  Implements a bounded code-change request in an isolated git worktree. Runs type-check + build after every logical unit of work, commits when clean, pushes its feature branch to the configured remote. Returns a 2-line summary. Spawned by the orchestrator during the sibling-feature workflow (the orche

• amwst-scenario-improvement-implementer

  Reads scenario_proposed-improvements_*.md files from reports/scenarios-runner/ and implements the P0 items in an isolated git worktree. Auto-detects the project's type-check and build commands (or reads them from tests/scenarios/scenarios.config.json). Commits each P0 item individually. Returns the

• amwst-scenario-proposer

  Produces the 11th-HOUR improvement PROPOSALS for ONE scenario run, in its own isolated forked context, AFTER the amwst-scenario-runner has finished. Reads the runner's structured report at reports/scenarios-runner/SCEN-NNN_<ts>.report.md plus the scenario file (greppably) and writes reports/scenario

• amwst-scenario-runner

  Executes ONE UI scenario end-to-end in its own isolated forked context. Reads the scenario file at ${CLAUDE_PROJECT_DIR}/tests/scenarios/SCEN-NNN_*.scen.md, follows the rules in SCENARIOS_TESTS_RULES.md, drives the app UI via the dev-browser plugin (loaded via the dev-browser:dev-browser skill — san

• web-scenario-tester-main-agent

  Browser-driven UI scenario tester (AI-Maestro MEMBER role). Drives a running web app's UI through dev-browser to AUTHOR, RUN, and BATCH end-to-end `.scen.md` scenario tests, with fix-as-you-go, screenshots, structured reports, and 11th-hour improvement proposals. Use when invoked as `claude --agent

Hooks (1)

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/scripts/amwst_subagent-write-guard.sh"

Similar plugins