julesmanson/clod-code-agentic-skills

A series of agentic developer skills for Claude Code, adaptable to other AI coding assistants with minimal edits.
## Contents
Skills
QA and QC
What is a skill?
Further reading
Repository structure
Installing a skill
Contributing
License
## Skills
Skill
Description
Commit
Commit and push to the appropriate GitHub repo (adaptable to other git services with minimal edits). Prompts before creating a new repo if one doesn't exist. Supports new repo [name] [visibility] [license] commit to create a repo and make the first commit in one step. Full usage, examples, defaults, and other-git-host notes: commit/README.md .
Autonomon
Personal, always-on workflow fixes for this repo's author, originally intended for his own personal use, now open to the public for anyone's convenience.
generate file — save generated output to a GENERATED folder instead of leaving it sandboxed
terse — keep responses short by default
verdict first — state the verdict before the explanation
im thinking — treat a stated impression as an impression, not a claim to cross-examine
ballpark — treat a bare figure as illustrative magnitude, not a precision claim
load bearing — in multi-point critique, lead with corrections that actually change the conclusion
memory cap — stay aware of a user-set ceiling on memory files, flagging before a new one would exceed it
Details: autonomon/README.md .
Thought Experiment
Runs a rigorous, evidence-based audit of any target — script, doc, config, whole system — by constructing concrete failure-scenario "thought experiments" and verifying every claim about a dependency's real behavior against a live source rather than memory. Triggered by the phrase "thought experiment." Details: thought-experiment/README.md .
Evaluate Me
Runs a strict, evidence-gated self-assessment of the user across nine points, then closes with a calibration note. Model-agnostic with minimal adaptation.
Effective use of an AI assistant — how well the user directs, delegates to, and collaborates with an AI assistant
Project-specific technical judgment — architecture, design patterns, testing/verification discipline, and how errors get resolved
General competence — technical skill and judgment beyond any one project
Work ethic — discipline, follow-through, and whether stated standards hold up under pressure or fatigue
Consistency between stated principles and actual behavior — whether the user's stated values show up in what they actually did
Personality — communication style, humor, self-awareness, and response to correction versus praise
Markers of higher intelligence — ten evidence-gated reasoning-quality signals (novel problem solving, abstraction, transfer, fluid reasoning, counterfactual reasoning, prediction/mental modeling, metacognition, cross-domain synthesis, compression, pattern learning)
Points that need improvement — specific, actionable patterns worth working on next, synthesized strictly from the seven points above
Overall tier — a holistic synthesis across all eight points above (Developing / Competent / Strong / Exceptional, not a letter grade or computed average), paired with a sentence naming which point or two drove the call