skills.rest
OpenSourceAGI
OpenSourceAGIOFFICIAL
Explore all skills by this author →
GitHub

acceptance

Verifies software deliveries end-to-end by capturing evidence and publishing immutable acceptance rounds.

74|11|Updated Jul 4, 2024
Search Tags:#screenshot#qa automation#acceptance testing#agent-browser#evidence capture#e2e verification#lh cli
One-click install
npx skills add https://github.com/OpenSourceAGI/qwksearch-research-agent --skill acceptance-opensourceagi
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: acceptance
Source: https://github.com/OpenSourceAGI/qwksearch-research-agent/tree/main/apps/qwk-in-lobe/packages/builtin-skills/src/acceptance
Command: npx skills add https://github.com/OpenSourceAGI/qwksearch-research-agent --skill acceptance-opensourceagi

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser, and includes references (resource) components.

What problem does it solve? Delivering code changes without proof they actually work forces reviewers to trust claims instead of inspecting results. This Skill drives the real product across CLI, web, Electron, native macOS, and iOS Simulator surfaces, captures visually confirmed evidence, and publishes a structured acceptance round that a human can accept or reject. ## Core Features & Use Cases - Plan discovery or authoring: Satisfy an existing verify plan when handed an operation id, or author your own human-judgeable checks and publish a standalone acceptance with the lh CLI. - Multi-surface evidence capture: Drive the correct proving surface (agent-browser for web/Electron, osascript for native macOS, simctl/AXe for iOS Simulator) and submit screenshots, DOM snapshots, text output, audio, video, and transcripts per criterion. - Immutable rounds with coverage self-check: Publish repair work as new rounds on the same acceptance, verify every requiredEvidence type landed, and hand off a single stable acceptance URL. - Use Case: After fixing a UI bug, run the app, screenshot the corrected state, submit it against the plan criterion, and deliver an acceptance page link the reviewer can inspect without running anything locally. ## Quick Start Verify my latest change end-to-end, capture evidence on the right surface, and publish an acceptance round with the lh CLI.

Frequently Asked Questions about acceptance

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify a code change with real evidence before delivery?▼

Discover the existing verify plan with lh verify plan state, or author your own checks, then drive the matching surface and submit one artifact per criterion using lh acceptance run result submit. Finish by confirming every requiredEvidence type is present and sharing the acceptance URL.

Which surface should I use to test web, Electron, or iOS changes?▼

Use CLI output for backend changes, agent-browser for web and Electron renderers, osascript with screencapture for native macOS, and simctl with an AX CLI for iOS Simulator. Pick the cheapest surface that can actually prove the criterion.

Do I need an operation id or LobeHub environment to run acceptance verification?▼

No operation id is required. Without one, you author the checks yourself and lh acceptance run ingest creates a standalone acceptance round; an operation id is only used when the invoking task explicitly hands one to you.

Can unit tests or lint results count as acceptance checks?▼

No. Programmatic gates like test suites, type-checks, lint, and coverage are explicitly forbidden as acceptance checks and are dropped at ingest. They may be reported as one narrative line, while checks must be outcomes a person judges.

How do I fix a failed acceptance round after changing the code?▼

Publish a new round rather than editing the old one, since rounds are immutable. Read the current acceptance with lh acceptance view, reuse stable check ids for rejected items, and pass the previous acceptance id so the new snapshot joins the same history.

Why is my screenshot evidence not enough to pass a criterion?▼

A criterion declaring requiredEvidence cannot pass on text alone; every declared evidence type must be uploaded or the check stays uncertain. Screenshots must also be opened and visually confirmed to show the expected state, not just captured.

Related Skills

Explore More Software Engineering Skills→
learntocloudlearntocloud
801

validate

Runs the repository quality gate and smoke-tests the FastAPI endpoints.

Official
Intermediate
harkersharkers

reviewer-agent

Review bounded work for correctness and output evidence-linked findings with a disposition.

Community
Advanced
NEXUZ-SYSNEXUZ-SYS
1

napkin

Read and curate the Napkin Runbook format at session start.

Official
Advanced
The13thNodeThe13thNode
3

test-automation

Generate unit, integration, and end-to-end test suites for JavaScript/TypeScript projects.

Community
Advanced