Skip to content

bioepic-data/ecosim-agent / .agents/skills/ecosim-data-readiness

ecosim-data-readiness

Verified

Inventory staged local EcoSIM data at project startup and validate input readiness for a selected site, simulation period, and preprocessing or analysis workflow. Use before data-dependent work or when relevant inputs change.

EcoSIM Data Readiness

Establish what local inputs are available and what has actually been validated for the requested workflow. Use two modes: a lightweight startup inventory and a focused validation before data-dependent work.

Startup inventory

  • Run once per session after the repository startup notice. Search ./data first, then the wider repository, including inputs, example_input, templates, and existing result folders when present. Honor explicitly supplied input paths as well.
  • Prefer rg --files and targeted filename searches. Check whether data is a symlink and whether its target is accessible; a missing or inaccessible root is a finding, not evidence that data do not exist elsewhere. Avoid following unrelated links or scanning external storage beyond staged data roots.
  • Identify candidate categories: site metadata; meteorology; soil profiles; plant traits/PFT parameters; plant and soil management; deposition or precipitation chemistry; greenhouse-gas forcing; and benchmark observations or model outputs. Record representative paths and apparent sites or periods.
  • Keep startup inexpensive: inspect paths and existing manifests, without opening every large dataset, extracting archives, or validating every site. Label coverage inferred from filenames as unverified. Templates and example files are candidates, not confirmed site-specific inputs.
  • If no site or task is selected, report inventory only. Do not declare global readiness or treat every category as required for unrelated code work.

Workflow validation

When preparing driving data, read references/site_conflicts.md. Check site evidence before extraction and again against generated inputs; record contradictions, defaults and unavailable evidence separately from format validity.

For soil texture or organic carbon, use the field-soil evidence review. Site BIF records may supply measurements absent from map products, but missing depths and conflicting stock estimates remain unresolved until their support and units are established. Passing the offline conflict tool is not this review.

For cropland, verify the user's maximum 3 m active soil-depth policy and a literature-supported maximum rooting layer using ../ameriflux-surgo-grid-extract/references/cropland_depth_policy.md. Check the runtime layer selectors as well as the CDPTH array; effective irrigation depth and total maximum rooting depth describe different quantities.

Determine requirements from the user's task, selected site and period, relevant workflow skill, configuration/namelist, and consuming reader or template. Check only dependencies of that workflow: for example, output analysis may require model output and observations without requiring all simulation input categories. If necessary context cannot be inferred, request the missing site, period, or input choice while continuing independent work.

For runnable site-case preparation, the mandatory whole-case preflight checks the actual namelist and selected files. Assembly and launch helpers enforce it. Re-run after input/namelist edits; stop on invalid or unsupported active inputs, and keep unresolved scientific evidence visible as warnings.

For an EcoSIM run namelist, use ecosim-climate-readiness for the automated climate-file check. It resolves forcing periods and restart scope, then checks actual required records. Preserve its errors and clock/provenance limitations; year-coordinate coverage alone is not climate readiness.

For each required input, select the actual file and inspect its contents:

  • Access and identity: Verify readability, format, required variables, dimensions, and site coordinates/ID. Distinguish archive members from usable extracted inputs. Record unresolved choices between conflicting versions.
  • Time: Verify requested coverage, cadence, calendar, timezone, timestamp meaning, duplicate or missing steps, and management-event alignment. Identify spin-up or cycling assumptions separately from observed coverage.
  • Units and values: Check documented units, fill values/sentinels, missing data and quality flags, and physical ranges appropriate to the variable. Distinguish flux rates from interval totals; verify conversions between mass and molar units, elemental and compound mass, and area/depth bases. Report potential mass-balance inconsistencies with the evidence available.
  • Compatibility: Check layer depths/order, grid/site dimensions, PFT codes, parameter identifiers, management selectors, and cross-file consistency against the actual EcoSIM reader or maintained workflow conventions. Confirm referenced namelist paths exist when preparing a runnable case.
  • Metadata: For NetCDF, inspect units, descriptive names, coordinates, calendar/time encoding, and declared CF conventions. Use standard names where applicable; do not invent CF names for EcoSIM-specific variables. Report metadata gaps separately from reader incompatibilities. Do not claim full CF compliance from a Conventions attribute alone.

Use headers and lazy reads first; inspect data arrays only as needed for the selected validation. Reuse existing domain validators where appropriate, reading their skill instructions when invoking them. Record unavailable tools or incomplete checks as unverified, rather than assuming success.

Findings and reports

Explain each issue using the actual data: identify the conflict type, compare values on explicit units/volume/depth/time bases, show the violated relationship, and describe its consequence and a concrete resolution. Separate confirmed contradictions, conditional scenarios and unavailable evidence. Follow the detailed reporting contract in references/site_conflicts.md; a code or label such as "density conflict" alone is insufficient. For quantitative conflicts, include a worked numerical substitution and affected-layer/date calculations sufficient to reproduce the result; follow the reproducible-calculation contract in that reference. A percentage or qualitative summary alone is insufficient. Density/macropore mass differences require demonstrated matching measurement pore domains to become errors; otherwise report conditional warnings. Use the soil-volume reference for measurement-method and PTF plausibility checks.

Assign each relevant input one status:

StatusMeaning
Found, unverifiedCandidate exists; contents or required checks remain unverified.
ReadyRequired checks passed for the stated site, period, and workflow.
Needs preparationSource exists but conversion, extraction, or another documented preparation step remains.
MissingNo candidate was found in the searched locations for a required input.
InvalidA required check failed; describe the failure and affected work.
Not requiredThis input category is outside the selected workflow's requirements.

Summarize startup findings briefly in chat. Save a detailed inventory or validation report under ./result/data_readiness/, using a site/task identifier and timestamp to avoid overwriting another assessment. Include mode, assessment time, site/task/period (or unspecified), searched locations, actual input paths, checks performed, evidence, assumptions, status, and concrete next steps. An inventory report must state that content validation has not been performed.

Overall readiness requires all required inputs to be ready for the declared scope; optional missing observations need not block a simulation. Explain which steps are blocked by missing, invalid, unprepared, or unverified dependencies, and proceed with independent work. A readiness assessment does not itself authorize downloads, credential access, changing source data, or running an expensive simulation.

Reuse the session's assessment until relevant file paths, sizes, modification times, configurations, or task requirements change. Recheck affected inputs after preparation or edits. Treat reused reports as dated evidence and verify their scope and input freshness before relying on them.

planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible.Stars: 118kUpdated Claude CodeCodexVerifiedInstall
screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe.Stars: 118kUpdated Claude CodeCodexVerifiedInstall
contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies.Stars: 108kUpdated Claude CodeCodexVerifiedInstall
meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要".Stars: 108kUpdated Claude CodeCodexVerifiedInstall