Assemble executable soft-mode recovery evidence for Minimum Stein Discrepancy Estimator proxy experiments.
Scanned 9/9/2026
Install to Claude Code
npx -y skills add VectorSpaceLab/AREX-Skill --skill msd_recovery_protocol --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Msd Recovery Protocol?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vectorspacelab-msd-recovery-protocol)More formats (shields.io, HTML) on the badges page.
---
name: msd_recovery_protocol
description: Assemble executable soft-mode recovery evidence for Minimum Stein Discrepancy Estimator proxy experiments.
---
# MSD Recovery Protocol
Use this skill when recovering Minimum Stein Discrepancy Estimators under bounded runtime. It coordinates the generated Stein-kernel, diffusion-design, and optimiser skills to produce validation-ready recovery artifacts without reading an original source repository.
## Inputs
- Attempt directory containing `module_plan.json`, module docs, generated skills, and `environment/runtime_handoff.json`.
- Recovery mode and target from the run manifest.
- Paths to generated skills that must be invoked or cross-checked.
## Outputs
- `recovery/experiment_plan.md` describing full and proxy options.
- `recovery/source_manifest.json` listing only allowed sources.
- `recovery/logs/generated_skill_invocations.json` with evidence for every core module.
- `recovery/logs/generated_data_item.json` and `recovery/logs/training_trace.json` for reduced proxy evidence.
- `recovery/recovery_result.json` and `recovery/experiment_validation.json`.
## Workflow
1. Confirm full reproduction blockers and soft-mode permission for a declared proxy.
2. Generate a deterministic Student-t location sample matching the module-plan target.
3. Call or import the generated diffusion-design skill to obtain Student-t diffusion diagnostics.
4. Call or import the Stein-kernel skill to evaluate DKSD losses.
5. Call or import the optimiser skill to minimise the empirical loss.
6. Save all logs and mechanism checks before running the recovery validator.
## Validation
Run the bundled `tests/test_msd_recovery_protocol.py` for source-manifest and mechanism-check helpers. The full recovery harness in the attempt directory should additionally run `validate_recovery_experiment.py`.
## Limitations
This skill accepts a declared reduced/proxy target only in soft mode. It must never label proxy evidence as full reproduction of the paper figures.
## Refinement cycle 3 note
A negative source-boundary check rejected an `original_repo` source; keep this check as a required recovery-protocol regression.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!