Aise review

PhysisAI: trusted, local-first personal and project memory for Codex and Claude Code.

Install

npx -y skills add jiayx01/PhysisAI --skill aise-review

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

18 days oldThe repository was created 18 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
1 stars1 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.

Copied from the file, not written here

审阅 PhysisAI 候选经验的来源、适用范围、重复项、冲突、时效、上下文成本和建议去向。用户要求看看学到了什么、候选是否正确、该写进哪里或为什么被阻止时使用；只审阅不晋升。

1.9 KB, 362 tokens by cl100k_base, as published. Nobody here has run it

Run physisai verify --root <project-root> before trusting state.
Run physisai review --root <project-root> [proposal-id].
Inspect the live source named by source_ref; do not accept a stale summary when the source is cheap to recheck.
Check:
- Would a capable model know this without memory?
- Is it stable and likely to recur?
- Is the source authoritative enough for the claim?
- Is global or project the narrowest correct scope?
- Should the item cross initialized projects in the user layer (global) or remain isolated in this project (project)?
- Does an existing memory already cover it?
- Does it contradict current instructions or active memory?
- Should it expire or supersede an older item?
- Is the context cost justified by avoided failure or repeated effort?
- Can it remain one concise memory, or does a complex stable procedure belong in a Skill with scripts or references?
Check whether required_checks match the source, risk, and real task. Prefer deterministic project verifiers when they exist; do not add a generic checklist merely to make the candidate look complete.

Return one recommendation:

test: evidence is adequate and required checks are clear;
edit: candidate is useful but content, scope, triggers, or target need correction;
reject: generic, volatile, unverified, duplicative, unsafe, or not worth context cost;
hold: evidence is still immature.

Show facts, inferences, and recommendation separately. Do not run promotion, rejection, rollback, migration, or mutate active files from this skill.

278 B alongside SKILL.md