aif-verify

Verify completed implementations against plans, build results, tests, and project rules.

31|2|Updated Mar 21, 2026
One-click install
npx skills add https://github.com/letuhao/lore-weave --skill aif-verify-letuhao
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: aif-verify
Source: https://github.com/letuhao/lore-weave/tree/main/.cursor/skills/aif-verify
Command: npx skills add https://github.com/letuhao/lore-weave --skill aif-verify-letuhao

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? After an AI-assisted implementation finishes, it is hard to know whether every planned task was actually completed, whether the code compiles and passes tests, and whether the work still respects the project's architecture, rules, and roadmap. This Skill performs a structured post-implementation audit so gaps are caught before commit or merge. ## Core Features & Use Cases - Task Completion Audit: Walks every task in the plan file and marks each as complete, partial, not found, or skipped by inspecting the actual code with Glob and Grep. - Quality Checks: Runs build, test, and lint commands matched to the detected stack (TypeScript, Go, Python, Rust, PHP) and checks dependencies, leftover TODO markers, and undocumented environment variables. - Context Gates: Evaluates architecture, rules, and roadmap alignment with normal and strict modes, then appends a machine-readable aif-gate-result JSON block for orchestrators. - Use Case: After running /aif-implement on a feature branch, invoke verification to confirm all 8 planned tasks exist in code, tests pass, and no undocumented SENDGRID_API_KEY-style config slipped in before committing. ## Quick Start Ask the assistant to verify the completed implementation against the current plan, optionally in strict mode before merging.

Frequently Asked Questions about aif-verify

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

FAQPage Schema
How do I verify an AI-generated implementation is complete?

Run the verification skill after implementation finishes. It reads the plan file, checks each task against the actual code using search tools, and reports each task as complete, partial, not found, or skipped with file-level evidence.

What does strict mode check before merging a branch?

Strict mode requires all tasks complete, build passing, tests passing, zero lint warnings, no leftover TODO markers, no undocumented environment variables, and passing architecture, rules, and roadmap gates. Invoke it with the --strict flag before merging.

Which build and test tools does the verification support?

It detects the stack from project files: tsc or npm build for TypeScript, go build and go test for Go, pytest for Python, cargo check and cargo test for Rust, and phpunit for PHP. Linting uses ESLint, golangci-lint, ruff, or php-cs-fixer when configured.

Can verification run without a plan file?

Yes. If no plan file is found, it offers standalone options: verify the last commit or verify the current branch diff against the configured base branch. Plan-based verification remains the primary path.

Does the verification skill modify project context files?

No. It is read-only for context artifacts like DESCRIPTION.md, ARCHITECTURE.md, ROADMAP.md, and RULES.md. When it detects drift, it reports the finding and suggests the owning command, such as /aif-architecture or /aif-roadmap, to fix it.