Skip to content
Skills Library
Science & research · Analysis

reduce

/reduce

End-to-end reduction of a CASA Measurement Set, raw data to a first image: inspect, pre-calibration, calibration, optional post-cal flagging and polarisation, imaging. Invoke when asked to take an MS through to an image.

Licence
GPL-3.0
Last repository change
20.10.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
2
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 20.10.2026, without references to retired models or APIs.

Licence

GPL-3.0, otwarta.

What it does and when to use it

End-to-end reduction of a CASA Measurement Set, raw data to a first image: inspect, pre-calibration, calibration, optional post-cal flagging and polarisation, imaging. Invoke when asked to take an MS through to an image.

Original author description

End-to-end reduction of a CASA Measurement Set, raw data to a first image: inspect, pre-calibration, calibration, optional post-cal flagging and polarisation, imaging. Invoke when asked to take an MS through to an image.

How it works

  1. The contract: stages.json
  2. Stage order
  3. Work off the tools when you need to, and record it
  4. At the end
Original SKILL.md headings · 59 lines, 478 words
SKILL.md preview
---
name: reduce
description: >
  End-to-end reduction of a CASA Measurement Set, raw data to a first image:
  inspect, pre-calibration, calibration, optional post-cal flagging and
  polarisation, imaging. Invoke when asked to take an MS through to an image.
allowed-tools: mcp__plugin_radio-analyst_ms-inspect__ms_observation_info,
               mcp__plugin_radio-analyst_ms-inspect__ms_field_list,
               mcp__plugin_radio-analyst_ms-inspect__ms_scan_list,
               mcp__plugin_radio-analyst_ms-inspect__ms_scan_intent_summary,
               mcp__plugin_radio-analyst_ms-inspect__ms_spectral_window_list,
               mcp__plugin_radio-analyst_ms-inspect__ms_correlator_config,
               mcp__plugin_radio-analyst_ms-inspect__ms_antenna_list,
               mcp__plugin_radio-analyst_ms-inspect__ms_baseline_lengths,
               mcp__plugin_radio-analyst_ms-inspect__ms_elevation_vs_time,
               mcp__plugin_radio-analyst_ms-inspect__ms_parallactic_angle_vs_time,
               mcp__plugin_radio-analyst_ms-inspect__ms_shadowing_report,
               mcp__plugin_radio-analyst_ms-inspect__ms_antenna_flag_fraction,
               mcp__plugin_radio-analyst_ms-inspect__ms_workflow_status,
               mcp__plugin_radio-analyst_ms-inspect__ms_verify_import,
               mcp__plugin_radio-analyst_ms-modify__ms_set_intents,
               mcp__plugin_radio-analyst_ms-modify__ms_apply_preflag,
               mcp__plugin_radio-analyst_ms-inspect__ms_online_flag_stats,
               mcp__plugin_radio-analyst_ms-inspect__ms_flag_summary,
               mcp__plugin_radio-analyst_ms-modify__ms_generate_priorcals,
               mcp__plugin_radio-analyst_ms-inspect__ms_verify_priorcals,
               mcp__plugin_radio-analyst_ms-modify__ms_setjy,
               mcp__plugin_radio-analyst_ms-inspect__ms_verify_model,
               mcp__plugin_radio-analyst_ms-modify__ms_initial_bandpass,
               mcp__plugin_radio-analyst_ms-inspect__ms_refant,
               mcp__plugin_radio-analyst_ms-inspect__ms_verify_caltables,
               mcp__plugin_radio-analyst_ms-inspect__ms_plot_caltable_library,
               mcp__plugin_radio-analyst_ms-modify__ms_apply_initial_rflag,
               mcp__plugin_radio-analyst_ms-inspect__ms_residual_stats,
               mcp__plugin_radio-analyst_ms-modify__ms_gaincal,
               mcp__plugin_radio-analyst_ms-inspect__ms_gaincal_snr_predict,
               mcp__plugin_radio-analyst_ms-inspect__ms_calsol_stats,
               mcp__plugin_radio-analyst_ms-modify__ms_bandpass,
               mcp__plugin_radio-analyst_ms-modify__ms_flag_caltable,
               mcp__plugin_radio-analyst_ms-modify__ms_fluxscale,
               mcp__plugin_radio-analyst_ms-modify__ms_applycal,
               mcp__plugin_radio-analyst_ms-modify__ms_postcal_flag,
               mcp__plugin_radio-analyst_ms-inspect__ms_pol_cal_conditions,
               mcp__plugin_radio-analyst_ms-modify__ms_setjy_polcal,
               mcp__plugin_radio-analyst_ms-modify__ms_polcal,
               mcp__plugin_radio-analyst_ms-modify__ms_tclean,
               mcp__plugin_radio-analyst_ms-inspect__ms_corrected_stats,
               mcp__plugin_radio-analyst_ms-inspect__ms_image_stats,
               mcp__plugin_radio-analyst_ms-create__ms_reduction_log,
               Bash,
               Read,
               Write,
               Edit
---

# Reduce: one MS from raw data to a first image

Load the `radio-interferometry` skill (`radio-analyst:radio-interferometry` when
installed as a plugin) first. It holds the science; this skill holds the order
and the checks. Read each of its stage files when you reach that stage.

## The contract: `stages.json`

`stages.json` is a sibling of this file. Read it now. It lists the stages in
order, and each stage's steps. A step is a `before` list of checks, an
`action` (a write tool), and an `after` list of checks. Its `source` names the
command step or skill section it comes from.

1. Run the `before` checks, then the action, then the `after` checks, before
   the next action. The checks are how you know the action worked; the skill
   file for the stage says how to read them.
2. Then append the call that worked to the ledger with `ms_reduction_log`,
   as 00-playbook.md "Record every step" says: one entry per write call, the
   tool's own name, literal params, and the `script_path` that ran.
3. Do not re-run a check whose earlier result is still valid. `cache` in
   `stages.json` says which actions invalidate which checks. Each MS is its own
   target: a split (`calibrators.ms`, `working.ms`) starts uncached.
4. `optional` stages (polcal, postcal_flag, selfcal) may be skipped. Say so in the report.
   Selfcal is never automatic: after the first image, assess it with 12-selfcal.md Step 0
   (`ms_image_stats` and `ms_residual_stats`), then decide, and report the decision and
   the numbers it rests on either way.

## Stage order

| Stage | Read | Command with the step detail |
|---|---|---|
| inspect | 01-workflow.md, 01b-workflow-phase2.md, 04-diagnostic-reasoning.md | inspect.md |
| precal | 10-precal-workflow.md | precal.md |
| calibrate | 07-calibration-execution.md | calibrate.md |
| postcal_flag (optional) | 13-postcal-rfi-flagging.md | none |
| polcal (optional) | 09-polcal-execution.md | polcal.md |
| image | 11-imaging.md | image.md |
| selfcal (optional, your decision) | 12-selfcal.md | none |

## Work off the tools when you need to, and record it

You may edit a generated script or run your own CASA code when the science
needs it. Then record it: a ledger entry with `script_path` set to the script
that ran (`tool='manual'` for your own code), and the change in `rationale`.
Read-only diagnostic scripts need no entry. The ledger must be enough to
replay the reduction: replay re-runs each recorded script in order. Measure the image you report
with `ms_image_stats` on the image the recorded step produced.

## At the end

When you stop, a hook scores your calls against `stages.json` and returns any
open items: a check not run, an action not called, a step not in the ledger,
a generated script edited without a record. Do them, or say in the report why
each was skipped. It does not block anything else.

Then build the report with the `report` skill (`radio-analyst:report` installed
as a plugin).
Files in this package (2)
  • SKILL.md 6 KB
  • stages.json 12 KB

Versions

RepositoryScoreLicence
This page · skunkworks-ra/radio-analyst100GPL-3.0
tkersey/dotfiles100MIT
alesaiani/agentflow100Apache-2.0
melodic-software/claude-code-plugins84MIT

Database snapshot: 13.11.2026 · package version af1a2be1ef09 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

No similar skills in the database.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.