All skills & plugins / Awesome Copilot skills (GitHub)
embedding-anisotropy-check
Use before reporting any cosine similarity computed over embeddings or transformer hidden states, whether the agent computed the vectors or loaded them. Measures the anisotropy of each side, centres by the population mean, and scores matched pairs against a permuted floor taken from the same comparison. Use when comparing one model's embeddings across precisions, runtimes or layers, when evaluatin
- Type
- Skill
- Catalog
- Awesome Copilot skills (GitHub)
- Source
- github/awesome-copilot/skills/embedding-anisotropy-check (3 files scanned)
- Install and details
- Agent Skills Index entry
- License
- MIT (SKILL.md frontmatter)
- Signals
- Bundled scriptsSpawns processes
Pre-approved tools
No allowed-tools or tools list in its SKILL.md, command or agent files, so the agent asks before using tools as usual.
Hooks
No hooks.
MCP servers and apps
None.
Bundled code
2 code files (Python):
embedding-anisotropy-check/scripts/example_scifact_halves.pyembedding-anisotropy-check/scripts/anisotropy_check.py
Network calls
No HTTP or socket call pattern found in the code.
Starts other programs
1 file. First match per file:
skills/embedding-anisotropy-check/scripts/example_scifact_halves.py:39res = subprocess.run([sys.executable, script, os.path.join(out, "A.npy"), os.path.join(out, "B.npy"), "--shared"],
Credentials referenced
No key, token or secret environment variable found in code or MCP config.
Static scan of commit 61fa1fa on 2026-11-02. Signals are patterns found in the files, not a verdict: a network call can be the whole point of a skill. Read the linked lines before you decide. How the scan works.