Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Code Reading — measured AI agent skill | DecimalAI

jamestorrevillas/code-reading

SkillScore82Maintained

Use this skill when navigating unfamiliar codebases, understanding legacy code, onboarding to a new project, or reverse-engineering how something works. Trigger on keywords: unfamiliar code, legacy code, understand this codebase, how does this work, new project, onboarding, read code, trace through, what does this do.

by jamestorrevillas

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Coding pack

decimalai skills pull jamestorrevillas-code-reading

Get this skill — curl · raw file

curl -fsSL https://app.decimal.ai/s/jamestorrevillas-code-reading/SKILL.md

Raw file: /s/jamestorrevillas-code-reading/SKILL.md ↗ · machine-readable summary at /s/jamestorrevillas-code-reading.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 6d ago

Yes — +5% lift+1% tokens0% turns82% of whole cases pass with skill (18/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-03	✗→✓	▲ Improved	-6%	0%
case-05	✓→✓	＝ Same ✓	10%	0%
case-02	✓→✓	＝ Same ✓	2%	0%
case-06	✓→✓	＝ Same ✓	11%	0%
case-08	✓→✓	＝ Same ✓	15%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

jamestorrevillas / dev-skills

.github/skills/code-reading/SKILL.md

View on GitHub ↗

License MITImported 8/1/2026Version v1

Skill contents

SKILL.md

Code Reading

Core Principle

Code is read 10x more than it's written. Reading code is a skill you actively develop — not a passive activity.

Codebase Onboarding Order

Always explore a new codebase in this order:

README — What does it do? How do I run it?
Package/dependency files — What tools and frameworks are used?
Entry point — Where does execution start? (main.ts, index.js, app.py)
Core domain models — What are the main data structures?
Key user flows — Trace one important feature end-to-end
Tests — Tests reveal intended behavior and edge cases
Config/env files — What is configurable? What are the environments?

Code Tracing Method

For understanding a specific flow:

1. Start from the trigger (user action, API call, scheduled job)
2. Follow the execution path step by step
3. Note: what data flows in? what comes out?
4. Identify where external systems are called
5. Mark where business logic lives vs. infrastructure

Questions to Ask While Reading

What is this component responsible for? (single responsibility check)
What does it need to run? (dependencies)
What does it produce/return?
What could make it fail?
Why was it written this way? (check git blame/history for context)

Using AI for Code Reading

"Explain what this function does, including:
- What it takes as input
- What it returns  
- Any side effects
- Non-obvious behavior I should know about"

"Walk me through how [feature] works, 
starting from [entry point] to [output]"

"What would I need to understand to safely 
modify [specific part] of this code?"

Navigating Large Codebases

Search patterns — use grep/ripgrep to find usages of a function/class
Git blame — who changed this and why?
Git log — when was this introduced? what changed over time?
Tests — run tests while reading to see what behavior is expected
Dependency graph — which modules depend on what?

Warning Signs While Reading

Functions longer than 50 lines (doing too much)
Deep nesting (> 3 levels) — logic is hard to follow
Many parameters (> 4) — likely needs refactoring
Comments explaining WHAT (should be obvious) vs WHY (actually useful)
No tests — higher risk, be extra careful when modifying

Want to see if jamestorrevillas/code-reading would activate on your phrasing? Try it before you install — no LLM call, no signup.

Other measured skills in the registry, with their headline benchmark lift.