FISKWISE / GETTING STARTED

Tell your agent what you want to do.

Fiskwise is a collection of independent, inspectable knowledge packs and R tools. Its install, update, and removal workflows are built into the agent instructions. Start with wise-writing and the shared R tools, then add data, presentations and methods when needed.

Preview changesKeep user-owned skillsReview scientific output

01 / CORE KNOWLEDGE

Install the shared foundation

Paste this prompt into an agent with local file and command access. It installs the core instructions first, checks the runtime, and asks before installing packages or system software.

Install Fiskwise

install https://github.com/DeepWaterIMR/fiskwise

You can also select the prompt and copy it with your keyboard.

YOUR FIRST TASK

Try a concrete question

Tell your agent what you have and what you want to produce. It should inspect the inputs, explain its approach and flag decisions that need your expertise.

Write a report

“Turn these R scripts into a reproducible Quarto report. Explain the methods and flag limitations.”

wise-writing →

Explore survey data

“Use BAIT to summarise catches by year. Check station keys, sampling effort and true zeros first.”

BAIT →

Build an assessment

“Review my catch series and survey index for a SPiCT assessment. Show diagnostics and sensitivity checks.”

spict-wise →

Present the findings

“Make a ten-minute presentation from this report. Propose a story and slide plan before building it.”

wise-presentations →

If setup is uncertain, ask “check Fiskwise”. Your agent will report what is ready and what needs attention.

02 / ADD ONLY WHEN NEEDED

Shared R tools, specialist packs on demand

Core skills install before R is needed. Runtime setup checks tidyverse, writing dependencies, ggOceanMaps and ggFishPlots. Package installation requires saved, revocable permission; system changes need separate approval. BAIT and modelling packs are optional.

R, tidyverse, and Quarto

Install these only when a project needs them and you approve. Quarto CLI is a separate application; the R package quarto is its R interface and does not replace the CLI.

Maps and life-history plots

ggOceanMaps and ggFishPlots are independent R packages. Fiskwise checks their GitHub versions during runtime setup and updates them with package permission. Additional map data and analysis choices remain task-specific.

IMR Biotic data

Install BAIT only for IMR Biotic work. Respect the data owner’s access and sharing conditions, especially for Russian and reference-fleet data. Follow BAIT privacy guidance and use the database path in its config unless you explicitly select another local database.

BAIT privacy guidance

Presentations

Add wise-presentations to turn a scientific report into a Quarto slide deck, starting with an agreed story and slide plan.

See an example →

Model workflows

Add index-wise, spict-wise or gadget-wise for a relevant analysis, with their stated pack and R-package prerequisites.

03 / MAINTAIN YOUR SETUP

Install, update, or remove Fiskwise

Copy a short command into a local coding agent: Claude Code, Codex, Antigravity, Mistral Vibe Code, Gemini CLI, Cursor, Copilot CLI, or OpenCode. These agents can follow the workflow when they have GitHub access and local filesystem and shell access; browser-only assistants cannot complete local setup. The install workflow sets up core packs first, checks R, Quarto and tidyverse, requests permission for package or system changes, and offers the add-ons your task needs. Removal asks you to confirm before deleting files. Agent compatibility details.

Update Fiskwise

update Fiskwise

You can also select the prompt and copy it with your keyboard.

Uninstall Fiskwise

uninstall Fiskwise

You can also select the prompt and copy it with your keyboard.

04 / SCOPE AND REVIEW

Research guidance, under researcher direction

Fiskwise describes scientific workflows and methods; each pack has its own domain and prerequisites. It is not a trained model, an institutional policy, or a substitute for project-specific protocols, applicable standards, or expert review.

Researchers direct and remain responsible for the work. Review methods, results and citations, and disclose AI use as required by the intended publication. ICMJE guidance on AI use.

Explore Fiskwise packs