How do I install the Project Recon skill?

Run npx skillmds@latest add microsoft/project-recon in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Project Recon safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Project Recon?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Project Recon free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Project Recon?

Microsoft (@microsoft) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Project Recon by microsoft — Agent Skill

Name: Project Recon
Author: Microsoft

Project Recon

Zero-dependency shell recon for any code repository — detect languages, count LOC, and report project scale. Pure POSIX find/wc or PowerShell, no Python or third-party tools required. Triggers: "how big is this project", "what languages", "project sizing", "repo recon", "LOC count", "scope check".

Microsoft 72de563 4 files · 8.5 KB Updated Sep 17, 2026 2.7k repo stars

File contents

Project Recon: Zero-Dependency Repo Sizing

What this skill does

Quickly answers: how big is this repo and what's in it. Shell-only (POSIX find + wc or PowerShell), runs anywhere, no install needed.

Output Format

Emit exactly this JSON structure — no additional fields:

{
  "total_loc": <int>,
  "languages": { "<lang>": <loc>, ... },
  "top_level_dirs": <int>,
  "primary_language": "<lang with highest LOC>",
  "git": <bool>
}

total_loc: sum of non-blank lines across all detected source files
languages: per-language LOC breakdown (only languages actually found)
top_level_dirs: count of immediate subdirectories under repo root (excluding hidden dirs)
primary_language: the language key with the highest LOC
git: whether the project root is a git repository (git rev-parse --git-dir succeeds)

Do NOT add fields beyond this schema. No descriptions, no module lists, no domain analysis.

Files

File	Purpose
`references/language-extensions.yaml`	Language → file extension mapping + generated-file suffixes to skip
`references/exclude-patterns.yaml`	Directory exclude rules (global + per-language)
`references/loc-shell.md`	bash + PowerShell counting templates

Workflow

Detect languages. Scan for manifest files (pom.xml, *.csproj, package.json, go.mod, etc.) or dominant extensions. Look up language-extensions.yaml for the canonical extension list.
Resolve excludes. Union exclude-patterns.yaml::global.dirs with per_language.<lang>.dirs for each detected language.
Count. Use the template from references/loc-shell.md. Run once per detected language, sum for total.
Emit JSON. Output the schema above. Nothing else.

Semantics

Counts non-blank lines (wc -l), comment-inclusive. This is intentional — speed and simplicity over precision.
Expect 5–15% upward bias vs comment-stripping counters on verbose languages (Java, C#). Acceptable for sizing.

---name: project-recondescription: Zero-dependency shell recon for any code repository — detect languages, count LOC, and report project scale. Pure POSIX find/wc or PowerShell, no Python or third-party tools required. Triggers: "how big is this project", "what languages", "project sizing", "repo recon", "LOC count", "scope check".---# Project Recon: Zero-Dependency Repo Sizing## What this skill doesQuickly answers: **how big is this repo and what's in it.** Shell-only (POSIX `find` + `wc` or PowerShell), runs anywhere, no install needed.## Output FormatEmit exactly this JSON structure — no additional fields:```json{ "total_loc": <int>, "languages": { "<lang>": <loc>, ... }, "top_level_dirs": <int>, "primary_language": "<lang with highest LOC>", "git": <bool>}```- `total_loc`: sum of non-blank lines across all detected source files- `languages`: per-language LOC breakdown (only languages actually found)- `top_level_dirs`: count of immediate subdirectories under repo root (excluding hidden dirs)- `primary_language`: the language key with the highest LOC- `git`: whether the project root is a git repository (`git rev-parse --git-dir` succeeds)Do NOT add fields beyond this schema. No descriptions, no module lists, no domain analysis.## Files| File | Purpose ||------|---------|| `references/language-extensions.yaml` | Language → file extension mapping + generated-file suffixes to skip || `references/exclude-patterns.yaml` | Directory exclude rules (global + per-language) || `references/loc-shell.md` | bash + PowerShell counting templates |## Workflow1. **Detect languages.** Scan for manifest files (pom.xml, *.csproj, package.json, go.mod, etc.) or dominant extensions. Look up `language-extensions.yaml` for the canonical extension list.2. **Resolve excludes.** Union `exclude-patterns.yaml::global.dirs` with `per_language.<lang>.dirs` for each detected language.3. **Count.** Use the template from `references/loc-shell.md`. Run once per detected language, sum for total.4. **Emit JSON.** Output the schema above. Nothing else.## Semantics- Counts non-blank lines (`wc -l`), comment-inclusive. This is intentional — speed and simplicity over precision.- Expect 5–15% upward bias vs comment-stripping counters on verbose languages (Java, C#). Acceptable for sizing.

Frequently asked questions