IOWarp

The Abaqus program: scripting, running and auditing Abaqus/Standard and Explicit packages, Tosca optimization runs, 3D views and report figures of meshes and results, the abaqus-topology views, and the abaqus_engineer subagent.

Inspect ADIOS2 BP4 and BP5 output through clio-kit's adios MCP server (list, variables, steps, attributes, one variable at one step), with the adios2 skill: directory layout, steps and variables, shape/start/count, global and local arrays, blocks, attributes, the Python FileReader and Stream APIs, bpls, engines and operators.

OPAL's APPL-CORE source plugin: the mission of the appl-core agent and the APPL-CORE export contract (export_version 6, other versions refused) for plant-phenotyping data from ORNL's phenotyping facility, read from each export's own files. General onboarding and phenotyping methods come from the data-onboarding and phenotyping plugins.

Validated interactive system maps and diagrams as standalone HTML from a typed JSON spec, for one of five diagram types (architecture, workflow, sequence, dataflow, lifecycle). Use when the user asks to visualize architecture, a workflow, a call sequence, a data pipeline, or a state machine, or to map this repository. Not for ad hoc drawings or slide art, and not for Mermaid output; use the artifact tool for plain Markdown.

Decides the engineering approach for an intent in an interactive session: investigates, proposes two or three genuinely different approaches with trade-offs, recommends, lets the user decide, and writes the decision doc. Not a task-by-task plan; use sprint-plan. Not a multi-perspective debate; use design-council. Not product intent; use product-intent. Not a typed implementation handoff; use tech-spec.

Searches arXiv, summarizes or compares papers, and builds a compact literature survey as citation-ready, source-linked paper cards. Prefer the Researcher shadow agent for noisy multi-paper retrieval.

Structural code search with ast-grep rules: finds code by shape when grep on text is too noisy, and never rewrites it. Not for plain string or filename lookups; use grep or find.

Turns a finished PRD or architecture doc into a ticket backlog of small stories with verifiable acceptance criteria, confirmed, then created as GitHub issues or in another configured tracker. Not for local sprint slicing into a SPRINT.md; use sprint-plan.

Inspect computational benchmarks, scaling configurations, and performance harnesses.

Proves merged work on the canonical base, inspects and removes associated worktrees through Git, deletes local branches safely, and audits surviving repository refs. Not for creating worktrees; use worktree-create. Not for integrating branches; use worktree-merge.

Materials characterization method: X-ray CT and porosity, surface roughness and optical profilometry, fractography, and cross-technique defect synthesis, with the characterization_expert subagent.

Verify citation keys, DOIs, and bibliography entries against source files in research software.

Develop Clio Coder inside its own repository: trace source contracts, implement bounded changes, coordinate sprint work, and preserve running-agent and session correctness. Use before changing Clio source, prompts, tools, or library; pair with clio-coder-test for validation.

Optional Clio Coder coding and research skills, including Materio procedures. Native Clio agents and fleets require Clio Coder.

Validate Clio Coder changes using focused source regressions, built CLI and ACP fixtures, skill/library checks, and isolated package or GUI gates. Use when developing or diagnosing Clio itself; select checks from the changed behavior and report evidence without claiming unrun coverage.

The clio-data agent's plugin: its mission, assess-data-quality (what a variable states, checks, comparisons, visual checks) and convert-scientific-data (the HDF5/ADIOS2 to Parquet dtype map, skip-or-ask rules, and a bundled converter that reads back and verifies rows, nulls, NaNs, dtypes and per-column checksums).

Search and read the connected CLIO service's conversations, message passages and workspace files, with real navigable source links.

Correct-by-construction TypeScript rules (errors as values, parse don't validate, illegal states unrepresentable, deep modules) for writing or refactoring TypeScript when the project has not pinned its own standards.

Connect research data sources and work with the National Data Platform and knowledge stores.

Writes a durable, redacted, reference-not-copy handoff document when a session winds down or context is about to be compacted or lost, so the next session or agent can continue. Not for orienting at the start of a session; use context-prime.

Orients a fresh session in a repository by loading the last handoff, git state, the project constitution, and active-work signals before acting. Not for writing the handoff; use context-handoff.

What every CLIO agent carries: operating guidelines, rendering, the task lifecycle, conversation runtime tools and the react and workflow flows.

Handles API keys, tokens, and facility credentials (kerberos tickets, globus and scheduler tokens, ssh agents, netrc) before a tool needs one, after an auth error, when a new secret is supplied, or when one may have leaked. Not for provider target configuration; use clio-coder auth and target settings.

Slices an existing plan, PRD, or milestone into an executable sprint of dependency-ordered vertical slices sized for one agent run each, with done-when verification per slice; never fabricates a plan. Not for deciding the approach; use architecture.

Scientific dataset curation bundle with curation skill, curator agent, and inspection prompt.

General habits for unfamiliar datasets: read the self-description, inventory and audit tables with bundled scripts (sentinels, empty and near-duplicate columns, flags against data, join keys, declared against actual counts), keep an experiment card in the workspace, audit in a child agent, and state evidence plainly.

Deep web research as a flow: a planner chooses research directions, repeatable web researcher subagents investigate them in parallel, an independent web-enabled critic returns a typed audit (verified claims with fetched source and quote, rejected claims, gaps), gaps are researched within bounded rounds, and a Markdown report is written only from verified claims.

Delegate work: to a copy of the delegating agent, to the generic read-only explore, planner and review roles, to one of the agent's defined subagents, or to any installed agent.

Convenes several expert perspectives that challenge each other on a design decision with real trade-offs before code is written; quick mode runs a single round. Not for a one-question-at-a-time interrogation of a plan; use plan-interview. Not for splitting implementation across workers; use dispatch directly.

Document production practice: edit the canonical source, preserve Office and OpenDocument packages, validate the native format, review the rendered result, and apply anchored artifact review comments.

CLIO's managed document toolchain: converters, fonts and bounded document inspection.

Every document format CLIO works with: PDF, Word, spreadsheets, presentations, LaTeX and OpenDocument.

Create, edit and review Word documents.

The earthscope agent's L2 source plugin: its mission, acquisition of EarthScope GNSS station catalogs and position time series from the National Data Platform (discovered, never assumed), the earthscope-stations A2UI catalog for picking stations on a map, and the typed workflow_state for acquisition. GNSS methods come from the geodesy plugin.

Pre-registers a performance study, numerical comparison, parameter sweep, or benchmark: thresholds, tolerances, environment pins, and verdict conditions locked into the validation contract before any measurement. Not for diagnosing a stalled bug; use scientific-debugging.

Builds, tests, and debugs Clio Coder api 2 extensions with declared capabilities and host-rendered views. Not for instruction-only skills; use skill-authoring.

Factorio's identity and research process: its mission, specialist roster and working practice; the coordination, framing, study-log, audit and evidence fan-out skills; and its research-process subagents (methodologist, virtual lab, evidence researcher with its leaf and critic lanes, simulation methodologist, independent reviewer).

Fatigue and failure method: regime and mechanism, high-cycle and multiaxial fatigue, critical-plane (Fatemi-Socie) and defect-driven life prediction, fracture mechanics and crack initiation, with the fatigue_failure_expert subagent.

Finite-element and topology-optimization method, independent of any program: formulating a study, meshing and convergence, boundary conditions, structural response, reading stress and strain, and stating and interpreting an optimization problem.

Turns something noticed mid-session into a tracker issue: captures evidence, dedups against existing issues, composes a labeled issue with acceptance criteria, confirms, and creates it via gh. Not for batch ticket creation from a PRD; use backlog. Not for diagnosing an existing issue; use fix-issue.

Finds published skills and prepares their Library installation for the operator when the user asks whether a capability exists as a skill. Not for authoring a new skill; use skill-authoring.

Resolves a tracker issue end to end: fetches it, diagnoses only when the issue does not name the code, fixes test-first, and self-reviews against the acceptance criteria, leaving a verified uncommitted change. Not for filing; use file-ticket. Not for committing or PRs; use ship.

The models this agent can run work on: list_models, usage (each account's windows, balance and spend) and refresh_provider_models.

Geocode places, filter points by radius and query geospatial features through clio-kit's geo MCP server.

GNSS geodesy methods for any network: position time series (frames, trajectory models with rate, offsets and seasonal terms, noise and rate uncertainty) with a bundled fit, station catalogs read without guessing columns, region resolution, and station-coverage comparison over any discovered catalog.

Stress-tests a plan, design, or idea through a phased one-question-at-a-time interview before code is written, ending in a compact decision log. Not for a multi-perspective debate; use design-council.

Open and inspect HDF5 files through clio-kit's hdf5 MCP server, with the HDF5 team's hdf5 skill: how to use the tools with bounded reads, and reference guides on layout, chunking, filters, cloud and parallel access, drivers and clio-core ingest.

Launches, drives, or inspects another agent or command in a Herdr pane, tab, or workspace, including a second Clio Coder instance; requires HERDR_ENV=1. Not for background work a plain shell or dispatch already covers.

A simulated bench apparatus whose governing law is drawn from a secret seed: controllable inputs with stated ranges, noisy readings of the time for ten cycles, and every reading's history by branch, recorded for the study's views.

Archived: Minimal smoke-test plugin for packaged plugin hook discovery, trust, and invocation evidence.

Author LaTeX documents and build them to PDF.

Manufacturing and materials science method, independent of any simulation program: materials, manufacturing processes, characterization, mechanical testing, fatigue and failure, and finite-element and topology-optimization method. Programs such as Abaqus or Morphorm are separate plugins an agent selects alongside it.

How a design becomes a part: process families (additive and conventional), additive process parameters and their defects, build orientation, manufacturability checks, post-processing and surface finishing, and the manufacturing_expert subagent.

Interactive codebase maps from verified repository structure, symbols and imports, rendered natively as standalone HTML.

Materials science method for metallic parts: processing-structure-property reasoning, metallic material classes, nickel superalloys as a family, microstructure description and measurement, heat treatment and material property claims, and the materials_scientist subagent.

Materio: materials research interviews, supplied-corpus synthesis, lab feasibility, task execution and verified paper handoff.

Archived: Minimal smoke-test plugin that packages a plugin-local MCP server for marketplace launch-semantics validation.

Mechanical testing method: test and fatigue-campaign design against named standards, tensile testing, load frame and instrumentation capability, and data acquisition, with the mechanical_testing_expert subagent.

Long-term memory of the user, their projects and how they work, plus recall of past sessions.

The Morphorm topology-optimization program: Cubit mesh, input deck, XMLGenerator, batch run and restarts, with bundled checks of the Exodus mesh, the generated XML settings and the optimizer diagnostics.

Search National Data Platform datasets and stage their resources through clio-kit's ndp MCP server.

Create and edit OpenDocument text, spreadsheets and presentations.

Flows the model can start as owned work: run_workflow and run_flow, plus the council and arena flows.

Load, profile, filter and summarize tabular data through clio-kit's pandas MCP server.

Structured prompt template for summarizing scientific papers and research methodologies.

Parquet through clio-kit's parquet MCP server (single-file summaries, slices, previews and aggregates) and the parquet skill: footer, row groups and statistics, physical and logical types, hive partitioning, bounded scans with pyarrow or DuckDB, and write options.

Read, review and produce PDF documents.

Plant-phenotyping methods for any source: design and identity checks with MIAPPE-style example schemas and a validator, size and growth traits, treatment and genotype response, fluorescence and spectral signals, and phenotyping reports.

Multi-agent coordination recipe for reviewing scientific data pipelines.

Stress-tests a plan, design, or idea through a phased one-question-at-a-time interview before code is written, ending in a compact decision log. Not for a multi-perspective debate; use design-council.

Render static plots (line, scatter, histogram, heatmap, time series) through clio-kit's plot MCP server.

Create, edit and review PowerPoint presentations.

Turns an idea into a product requirements document through a phase-gated interview, ending in PRD.md plus per-milestone prompt files ready to drive a coding agent. Not for the problem-first product thesis; use product-intent.

Writes a problem-first product document for a greenfield effort: interviews for the problem, evidence, and a falsifiable hypothesis, with zero engineering decisions. Not for engineering decisions; use architecture. Not for turning a locked idea into milestone build prompts; use prd.

Answers a design question with throwaway code: a clearly marked, runnable, no-persistence prototype whose verdict is kept and whose code is discarded. Not for pre-registered performance experiments; use experiment-protocol. Not for production implementation.

Resolves a merge, rebase, or cherry-pick stopped on conflicts by reconstructing both sides' intent, validating with the project's own checks, and completing the operation. Not for integrating many branches; use worktree-merge.

A study's questions, hypotheses, experiments, observations, evidence, reframes and conclusions as typed, linked log entries, with sources read from the recorded calls, measurements as data, and the run-a-study method.

Diagnoses stalled or cross-system failures and wrong, NaN, nondeterministic, or unexpectedly slow scientific results through falsifiable hypotheses across distinct fault classes, with evidence-cited verdicts before any fix. Not for designing benchmarks or pre-registered experiments; use experiment-protocol.

General scientific practice for every science agent: method choice and study design, literature search and review, research gaps, citations, data analysis, statistics, uncertainty, figures, scientific writing, and non-interactive task runs.

Modernizes, ports, rewrites, packages, or replaces established scientific software in independently validated stages against an external scientific oracle, settling compatibility and upstream stewardship before calling it complete. Not for an isolated benchmark; use experiment-protocol. Not for diagnosing wrong results; use scientific-debugging.

A shell in the session workspace, with CLIO's managed execution runtime.

Ships finished work: writes one reviewed atomic commit, keeps maintainer branches local, or pushes a contributor branch to their fork and opens a PR only on explicit intent. Not for producing the change; use fix-issue.

Writes, reviews, or prunes a SKILL.md, judging whether its description, body, and length earn their cost. Not for packaging a workflow that just happened; use workflow-capture.

Writes, reviews, or prunes a SKILL.md, judging whether its description, body, and length earn their cost. Not for packaging a workflow that just happened; use workflow-distiller.

Slurm batch jobs: submit, record job ids, wait in bounded low-output steps with a bundled helper, and read a failed job's output before resubmitting.

Submits a batch script to Slurm through the clio-kit Slurm MCP server, polls it to a terminal state, fetches its output, and stamps the scheduler job id into the final answer. Not for writing the science inside the script or for interactive allocations.

SPOTTER's forensic watcher: provider-aware execution and artifact lineage across Flowcept, CMF and CLIO's native journal, captured-attention inspection, and skills for investigating and alerting on unexplained deviations in a watched CLIO session.

Slices an existing plan, PRD, or milestone into an executable sprint of dependency-ordered vertical slices sized for one agent run each, with done-when verification per slice; never fabricates a plan. Not for deciding the approach; use architecture.

Run commands on a configured remote host over SSH.

Actual model qualification over owned task inputs, with a defined worker subagent.

Isolated live model qualification of production task owners and controls.

Builds a feature or fix test-first, red to green at pre-agreed public seams, one vertical slice at a time. Not for designing benchmark criteria; use experiment-protocol.

Writes a typed call-stack architecture handoff: code-shaped contracts plus execution flows, implementation-ready for another engineer. User-invoked only. Not for weighing approaches or deciding the design; use architecture first.

Live-verification stress agent: the web MCP server and the synthetic v2 exerciser in one turn.

Live-verification agent exposing the full synthetic v2 exerciser tool matrix for leg C2.

Live-verification agent for the synthetic v2 MCP exerciser (tests/test_tools/mcp_exerciser.py).

Interactive analysis, dashboards, diagrams, 3D geometry for the mesh viewport (geometry-to-glb) and rendered visual review in CLIO's side panel.

Weather forecasts for a place.

Fetch web pages and search the web through clio-kit's web MCP server.

Live-verification agent for the web MCP server (web_fetch, web_search, web_fetch_events).

A Wordle bench for studies of solving strategy: a generated, versioned list of pronounceable five-letter pseudo-words, games whose answers are drawn from a secret seed the agent never sees, strict per-guess feedback, and a board record of every game for the study's views.

Packages a workflow that just happened into a reusable SKILL.md: reconstructs it from the session record, checks overlap with installed skills, gates on approval, then writes it following skill-authoring. Not for authoring a skill with no prior workflow; use skill-authoring.

Packages a workflow that just happened into a reusable SKILL.md: reconstructs it from the session record, checks overlap with installed skills, gates on approval, then writes it following skill-craft. Not for authoring a skill with no prior workflow; use skill-craft.

Read, search and edit files in the session workspace, and view images.

Stands up one or more git worktrees for parallel work, each on its own branch with gitignored config copied in, dependencies installed, and a verified health check. Not for merging finished worktrees; use worktree-merge. Not for closeout; use branch-closeout.

Integrates finished worktree branches through one throwaway integration branch, testing after each merge and running the full suite before the main line moves, with exact rollback on failure. Not for creating worktrees; use worktree-create. Not for closeout; use branch-closeout.

Create, edit and analyze spreadsheets and CSV/TSV tables.