Use this agent when a parent agent explicitly instructs it to audit the documentation quality and coverage of a Julia package module. This agent should NOT be invoked proactively — it runs only on direct instruction from a coordinating agent.\n\n \nContext: A parent agent is orchestrating a documentation improvement…
Use this agent when directly instructed or when you need a comprehensive, structured inventory of a Julia package's public API — including all exported or public functions, macros, and types, their full method signatures. This is typically the first step before running a style/convention audit, generating…
Use this agent when directly instructed or when you need a deep structural understanding of a Julia package's purpose, types, and public API without manually reading all source files. This is ideal before code reviews, architectural discussions, onboarding to a new codebase, or generating documentation outlines.\n\n…
Triage and resolve one item from a Julia package's issue tracker, ending in a code fix, a doc fix, or a reasoned "no change needed". Use when the user points at an issue number or link, says "look at issue N", or pastes an issue body and asks what to do about it.
Audit and improve Julia package docstrings: find missing/outdated docs on exported and public symbols, using Base.Docs.meta for context-efficient gap detection.
Audit a Julia package's array functions for 1-based indexing assumptions, add OffsetArrays/view tests that enforce the generic-indexing contract, and report fixes.
Use this skill when the user wants to implement the next step of a planned analysis project. This skill reads the plan, implements exactly one chunk, updates the plan, and prepares a clean handoff for the next session. Only use this skill when an ANALYSISPLAN.md exists in the project and the user wants to make…
Use this skill when the user wants to start a new analysis project and has an AGENTINSTRUCTIONS or project description to work from. This skill produces a persistent plan document that drives the companion /new-analysis-implement skill.
Rewrite a package's prose (docstrings, comments, docs pages, README) so it states what the code is and does now, in plain language, at the length the reader needs. Use when the user asks to clean up, simplify, or de-jargon comments or docs, complains that prose reads like a plan or a pitch, or wants history and…
Profile Julia code to find performance bottlenecks — CPU hot spots, runtime dispatch, GC pressure, allocations, and type instabilities — using a fully headless (no-GUI) workflow. Use whenever the user asks why Julia code is slow, wants it faster, mentions allocations or type instability, or asks to profile or…
Implement the next chunk of a planned design, API, or integration review. Reads DESIGNREVIEWPLAN.md, APIREVIEWPLAN.md, or INTEGRATIONREVIEWPLAN.md, implements one chunk (with a narrow batch exception for decide chunks), updates the plan, and prepares a clean handoff for the next session.
Update a Julia package's [compat] to allow a new breaking version of a dependency, then fix whatever the upgrade breaks. Use when the user says a dependency has a new major/breaking release, asks to "bump compat", "allow version X of Y", or a CompatHelper PR needs finishing by hand.
★not rated 71▲
+6
changed 1mo agoA65 tokens
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: