Skip to content

chore: sync no-external-llm hook #272

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
frecar wants to merge 1 commit into from
Closed

chore: sync no-external-llm hook #272

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 10 additions & 12 deletions scripts/check_no_external_llm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,12 +1,11 @@
#!/usr/bin/env python3
"""Enforce a configured LLM endpoint policy: no pay-per-token external
"""Enforce the cluster LLM policy (asgard#801): no pay-per-token external
LLM API endpoints in committed code.

This project is designed to be pointed at any OpenAI-compatible LLM
gateway (local Ollama, vLLM, llama.cpp server, hosted endpoint, ...).
Pay-per-token external APIs (`api.anthropic.com`, `api.openai.com`, etc.)
are forbidden in committed code paths to keep cost, privacy exposure,
and third-party dependency out of the runtime hot path.
The cluster's design target is `llm.carlsen.io` for all LLM calls. External
pay-per-token APIs (`api.anthropic.com`, `api.openai.com`, etc.) add cost,
privacy exposure, and third-party dependency. Documented in
`feedback_llm_local.md` and the policy issue.

This script flags any reference to those endpoints in committed files.
Test fixtures and benchmark scripts are exempt (they may legitimately
Expand Down Expand Up @@ -129,15 +128,14 @@ def main(argv: list[str]) -> int:

if violation_count:
print()
print(f"FAIL: {violation_count} violation(s) of configured LLM endpoint policy.")
print(f"FAIL: {violation_count} violation(s) of cluster LLM policy (asgard#801).")
print()
print("Project policy: route LLM calls through your configured gateway")
print("(set via LLM_API_URL). Pay-per-token external APIs")
print("(api.anthropic.com, api.openai.com, etc.) are forbidden in")
print("committed code paths.")
print("Cluster policy: route all LLM calls through llm.carlsen.io.")
print("Pay-per-token external APIs (api.anthropic.com, api.openai.com,")
print("etc.) are forbidden in committed code.")
print()
print("Fix options:")
print(" - Replace the endpoint with an env-driven default (LLM_API_URL)")
print(" - Replace the endpoint with https://llm.carlsen.io/v1 (or env-driven default)")
print(" - If the line is documentation that needs to name the forbidden")
print(f" pattern, append a trailing comment: `# {IGNORE_MARKER}`")
print(" - If it's a test fixture, move the file under tests/ (allowlisted)")
Expand Down
Loading