🤖 A collection of reusable agent skills, tools, and workflows designed to extend LLM capabilities and enable autonomous task execution.
Skill Claude CodeCodex
Generates Conventional Commit messages from staged changes in any git repository, and can auto-stage + group unstaged changes into a sequence of commits. Use when asked to write a commit message, generate a commit, draft a commit for staged changes, commit the current work, or when the user says "commit", "commit…
Skill Claude CodeCodex
Creates Architecture Decision Records (ADRs) — concise, durable documents that capture the context, decision, and consequences of significant architectural choices so future team members understand why things are the way they are. TRIGGER when: the user asks to create or write an ADR, document a decision, record why…
Skill Claude CodeCodex
Creates structured Request for Comments (RFC) documents that propose a change, compare alternatives against explicit decision criteria, and drive aligned decisions across teams. TRIGGER when: the user asks to create or write an RFC, draft a proposal, compare options, align stakeholders, get approval for a significant…
Skill Claude CodeCodex needs its repo
Root-causes complex bugs through a structured loop and saves a verified fix proposal — it diagnoses, it never fixes. Builds a fast deterministic feedback loop, reproduces, then probes ranked falsifiable hypotheses with parallel subagents and confirms the cause adversarially before writing the report. Use when a bug…
Skill Claude CodeCodex needs its repo
Spec-driven development workflow that takes a change from problem to shipped, verified code in four phases: specify, design, plan, execute — plus a fix command to correct course mid-stream. Auto-sizes from one-line fixes to multi-repo refactors. Invoke with /forge specify|design|plan|execute|fix. TRIGGER when: the…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: