agentsclimarketplace

Astrophotography processing

Skill mickpletcher/AI-Skills/claude/skills/astrophotography-processing

A portable library of reusable AI skills, prompts, workflows, schemas, and automation patterns designed for ChatGPT, Claude, Copilot, and local LLM workflows.

Install
npx -y skills add mickpletcher/AI-Skills --skill astrophotography-processing

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 13 stars13 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Router for astrophotography processing, troubleshooting, and safe workflow guidance across deep-sky, narrowband, planetary/lunar/solar, Milky Way landscape, comet, and mosaic data.

SKILL.md

7.0 KB, as published. Nobody here has run it

Astrophotography Processing

Purpose

Help users produce technically sound astrophotography results while preserving original data, metadata, and reproducibility.

Invocation Boundary

Use this skill only when context clearly indicates astrophotography or night-sky imaging workflows.

Trigger examples:

  • Telescope/lens star fields, deep-sky objects, Milky Way, comet, lunar/planetary/solar data
  • Calibration frames, stacking, registration, gradient removal, color calibration
  • FITS/XISF/RAW subs, linear masters, rejection maps, dithering, deconvolution
  • PixInsight, Siril, Astro Pixel Processor, DeepSkyStacker, StarNet, StarXTerminator, WinJUPOS, Python astro pipelines

When not to use:

  • General photo editing requests without astronomy context
  • Fashion, portrait, product, wedding, travel edits not involving night-sky capture workflows

Operating Mode

Choose one primary mode and label it in the response:

  • Diagnosis: inspect visible or file-based issues and identify likely causes.
  • Workflow: provide ordered processing steps for available data and software.
  • Recovery: repair a specific artifact or failed stage with minimal rework.
  • Automation: provide scripts, batch commands, or repeatable organization.
  • Scientific preservation: preserve measurement integrity and metadata over cosmetic output.

Evidence Rules

  • Separate observed issues from likely causes.
  • If only JPEG/PNG/screenshot is available, do not make firm claims about calibration quality, original bit depth, Bayer pattern correctness, clipping in source linear data, or linear behavior.
  • Prefer original linear masters, calibrated subs, FITS/XISF headers, and processing history before recommending irreversible fixes.
  • If files are unavailable, provide a cautious diagnosis and the smallest next test that can confirm/refute the hypothesis.

Minimal Questions Policy

Keep intake lightweight.

  • Ask at most 3 high-value questions first when critical context is missing.
  • If the user wants immediate guidance, proceed with reasonable assumptions and label assumptions explicitly.
  • Do not dump the full intake checklist unless requested.

High-value intake checklist (internal use):

  1. Image type: deep-sky, planetary/lunar/solar, comet, Milky Way landscape, mosaic.
  2. Input stage: raw subs, calibrated/registered subs, linear master, stretched image.
  3. Data/software: FITS/XISF/RAW/TIFF/SER and PixInsight/Siril/APP/DSS/Photoshop/Lightroom/GIMP/Affinity/Python.
  4. Capture and calibration basics: camera/optics/filter, exposure/gain/ISO/temp, darks/flats/bias/dark-flats.
  5. Desired output: aesthetic presentation, troubleshooting, or scientific/photometric integrity.

Local File Inspection Protocol

When local files are available:

  1. Inspect directory layout before proposing exact steps.
  2. Identify file types and separate lights, calibration frames, masters, and exports.
  3. Read metadata where possible: exposure, gain/ISO, temperature, filter, Bayer pattern, dimensions, bit depth, date/session, object.
  4. Summarize findings before recommending processing.
  5. Never modify source files.
  6. Write previews, audits, and generated outputs to process/ or exports/.
  7. If tooling is unavailable, fall back to filenames, folder structure, image dimensions, and user notes.

Helper scripts:

  • scripts/fits_audit.py
  • scripts/image_stats_preview.py

Core Workflow Router

Default deep-sky sequence unless software requires a variant:

  1. Organize data by session/filter/exposure/gain-temp.
  2. Calibrate with matched darks/flats/bias-or-dark-flats.
  3. Quality control and reject poor subframes.
  4. Register with stable reference geometry.
  5. Normalize/integrate with robust rejection.
  6. Linear stage: crop, gradient removal, color calibration, linear denoise, optional decon/sharpen.
  7. Stretch while protecting black point and star cores.
  8. Nonlinear finishing with restrained contrast/saturation/star handling.

Detailed decision guides:

  • Calibration details: references/calibration.md
  • PixInsight: references/pixinsight.md
  • Siril: references/siril.md
  • APP, DSS, Photoshop, Lightroom, GIMP, Affinity: references/app-dss-photoshop.md
  • Planetary/lunar/solar: references/planetary-lunar-solar.md
  • Milky Way landscape: references/milky-way-landscape.md
  • Comets, mosaics, OSC dual-narrowband: references/comets-mosaics-dualband.md
  • Artifact recovery: references/troubleshooting.md

Tool-Specific Guardrails

  • Do not use SCNR as a substitute for proper color calibration.
  • Apply green cast/noise suppression only after color calibration and only when appropriate.
  • Do not use Lightroom/ACR as the raw converter for calibration subframes unless this is a landscape workflow.
  • Optional WinJUPOS/de-rotation may help Jupiter/Saturn/Mars workflows.
  • Avoid presenting fixed numeric settings as universal; use adjustable starting ranges with visual/statistical checks.

Scientific or Photometric Mode

When the goal is analysis/measurement:

  • Preserve linear calibrated data.
  • Avoid cosmetic star removal, generative edits, aggressive denoising, star reduction, local contrast boosts, or nonlinear color manipulation in the measurement version.
  • Preserve FITS headers, WCS/plate-solve info, exposure metadata, and processing logs.
  • Export a presentation copy separately from the analysis master.

Solar Safety

For solar data, do not provide capture advice that implies viewing or imaging the Sun without a proper front-mounted solar filter or a purpose-built solar telescope. Processing guidance is allowed; unsafe capture assumptions must be corrected briefly.

Export and File Safety

  • Never overwrite RAW/FITS/XISF/SER/AVI/MOV/original TIFF sources.
  • Preserve linear masters in 32-bit FITS/TIFF or equivalent.
  • Use 16-bit TIFF for nonlinear finishing.
  • Use JPEG/PNG only for sharing.
  • Encourage use of templates/processing-log.md for reproducibility.

Default Response Structure

For normal requests, respond in this order:

  1. Likely issue or recommended path
  2. Assumptions based on available data
  3. Ordered workflow
  4. Checks after each major stage
  5. What not to do
  6. Suggested filenames/outputs

For shared images, respond in this order:

  1. Visible issues
  2. Likely causes
  3. Tests to confirm
  4. Smallest next edit
  5. Longer reprocessing recommendation (if needed)

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.