Gather context

Skill justincordova/agents/skills/gather-context

A structured AI coding workflow with skills, agents, and commands that make AI think before it builds.

Install

npx -y skills add justincordova/agents --skill gather-context

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when the user wants to quickly understand a codebase or project before doing work. Scans structure, conventions, dependencies, and patterns. Produces a concise summary, not files.

SKILL.md

3.1 KB, as published. Nobody here has run it

Quickly understand a codebase by scanning its structure, conventions, dependencies, and patterns. The output is a concise summary delivered in chat — no files created.

Announce at start: "Using gather-context skill to scan the project."

When to Use

Starting work on an unfamiliar codebase
Coming back to a project after a break
Before brainstorming or planning, when context is needed fast
When the user says "what's going on in this project?" or "catch me up"

Checklist

Create a task for each item and complete them in order. Run scans in parallel where possible.

1. Project identity

Find the package manifest (package.json, go.mod, Cargo.toml, pyproject.toml, .csproj, Directory.Build.props, etc.)
Identify: name, language, framework, runtime version
Read README.md or docs/SPEC.md for the project description

2. Structure

List top-level directories and files
Drill into each major directory to understand purpose
Note entry points (main.go, src/index.ts, cmd/, apps/, packages/)
Identify monorepo layout if present (workspaces, packages, apps)

3. Commands

Check package.json scripts, Makefile, justfile, Taskfile.yml, .csproj targets, or equivalent
Identify: dev server, build, test, lint, format, typecheck
Note the pre-commit gate (which commands must pass)

4. Dependencies

Scan runtime and dev dependencies
Identify the key libraries the project relies on
Note version constraints or pinned versions

5. Conventions and patterns

Read AGENTS.md or equivalent config if it exists
Check linter/formatter configs
Identify coding patterns: state management, auth flow, data layer, error handling
Note any non-obvious conventions (naming, file organization, import style)

6. Recent activity

Check recent git log (last 5-10 commits) to understand what's been happening
Check for uncommitted changes

Output Format

Present a concise summary in chat. Keep it scannable.

## Context: <project name>

**What it is:** <1 sentence>
**Stack:** <language, framework, key deps>
**Structure:** <brief layout description>

**Commands:**
- `cmd` — what it does

**Key patterns:**
- <pattern> — <where it lives>

**Recent activity:**
- <last few commits or active work>

If the user asked about something specific, focus the summary on that area. Skip sections that aren't relevant to what they're about to do.

Key Principles

Fast, not exhaustive. This is a quick scan, not a deep audit. Target 2-3 minutes.
Focused on what matters next. If the user is about to build a feature, emphasize the files and patterns they'll touch. If they're just exploring, give the broad picture.
No files created. The output is a chat summary. If the user wants persistent context, that's what the sync-agents skill is for.
Skippable. If context is already clear from the conversation, say so and skip the scan.