assess-it

Audits a repository with parallel review agents and files verified findings as GitHub Issues under an Epic.

Updated Mar 26, 2026
One-click install
npx skills add https://github.com/Sassy-Dog/sassydog-skills --skill assess-it-sassy-dog
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: assess-it
Source: https://github.com/Sassy-Dog/sassydog-skills/tree/main/skills/assess-it
Command: npx skills add https://github.com/Sassy-Dog/sassydog-skills --skill assess-it-sassy-dog

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Turning a whole-repository health check into actionable work is slow and error-prone: findings lack evidence, duplicates flood the tracker, and unaudited areas silently look clean. This Skill runs an evidence-based engineering audit and converts verified findings into a deduplicated, PR-sized GitHub Issue backlog under one tracking Epic. ## Core Features & Use Cases - Parallel multi-domain review: Fans out specialized review agents (architecture, security, testing, CI/CD, infra, observability, DX, dependencies) matched to the detected stack, each returning structured findings with mandatory file:line evidence. - Adversarial verification and dedupe: Every finding is checked against its cited code, deduplicated against existing open and closed GitHub Issues, and high-impact items can be challenged by skeptic subagents before filing. - Preview-gated filing with coverage ledger: Nothing is filed without user approval, and every run reports which domains were audited versus dark, both in the preview and inside the Epic body. - Use Case: Ask to assess a repo after a new model release; the Skill detects the stack, dispatches reviewers, verifies findings, aligns labels via the plugin's align-labels.sh script, and files child issues linked as native sub-issues of an executive-summary Epic. ## Quick Start Ask the agent to assess this repository and turn the verified findings into a GitHub Issue backlog under one Epic.

Frequently Asked Questions about assess-it

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

FAQPage Schema
How do I audit a whole repository and file GitHub Issues automatically?

Run the assess-it skill against the repo: it detects the stack, dispatches parallel review agents per domain, verifies each finding's file:line evidence, dedupes against existing issues, and files approved findings as child issues under one Epic.

How does the assessment avoid filing duplicate GitHub issues?

It fetches up to 500 open and 200 recently closed issues in Phase 0 as a dedupe index, matches findings by title and file area, and comments on matching issues instead of creating duplicates. Dedupe is re-checked per issue right before creation.

Does the repository audit create issues without asking first?

No. Filing is preview-gated: the skill prints the full Epic and every child issue plus a domain coverage ledger, then waits for explicit user approval before creating anything on GitHub.

What happens when a review agent fails to return results?

The domain is recorded as dark with its outcome (no report or could not dispatch) in the coverage ledger. Dark domains are never scored or called clean, the run continues, and the Epic body permanently records the gap.

Can I use this audit on any GitHub repository?

Yes, it is repo-agnostic and operates on one repo per run, defaulting to the current working directory. It requires the gh CLI authenticated with repo scope and dispatches only the review agents relevant to the detected stack.