[Mo0nl19ht]

What does this PR do?

get_prompt documents cache_ttl_seconds=0 as disabling caching, but it only skips the cache read. The fetched prompt is still written to the cache with a zero TTL, so a TTL-0 call:

• replaces the entry that default-TTL callers cached for the same prompt, and
• makes the next default-TTL call return that entry instead of fetching (as fresh within the same second, as stale with a background refresh after that).

_fetch_prompt_and_update_cache now skips the cache write when the TTL is 0. Eviction on a 404 is unchanged, so the cache still drops prompts the server no longer has. The LANGFUSE_PROMPT_CACHE_DEFAULT_TTL_SECONDS default path is not changed.

Type of change

• Bug fix
• New feature
• Breaking change
• Refactor
• Documentation update
• Tooling, CI, or repo maintenance

Verification

List the main commands you ran:

uv run --frozen ruff check .                                  # All checks passed!
uv run --frozen ruff format --check .                         # 595 files already formatted
uv run --frozen mypy langfuse --no-error-summary              # no issues
uv run --frozen pytest -n auto --dist worksteal tests/unit    # 1016 passed, 2 skipped

The new tests/unit/test_prompt.py::test_ttl_zero_fetch_does_not_write_to_cache covers an empty cache, the following default-TTL call, and an existing entry. It fails on main (the TTL-0 fetch leaves a PromptCacheItem in the cache) and passes with this change. E2E and live-provider suites were not run; the change needs no server.

Checklist

• I self-reviewed the diff using code_review.md.
• I added or updated tests for behavior changes.
• I updated docs, examples, or .env.template if needed. (Not needed; the docstring already describes this behavior.)
• I did not hand-edit generated files; if generated files changed, I used the upstream regeneration path.
• I did not commit secrets or credentials.

Confidence Score: 5/5

The PR appears safe to merge.

Summary

Zero-TTL calls now fetch prompts without adding or replacing cached entries.

• TTL-zero prompt fetches leave the shared cache unchanged.

_{Reviews (5) · Last reviewed commit: "fix(prompts): skip cache write when cach..." · Reviewed by Greptile}

[claude]

Claude Code Review

This pull request is from a fork — automated review is disabled. A repository maintainer can comment @claude review to run a one-time review.

[CLAassistant]

All committers have signed the CLA.

[chrikrah]

@Mo0nl19ht this matches the docstring ("Disables caching if set to 0"), and from my side it can be merged. The new test fails with only client.py reverted:

$ uv run --frozen pytest tests/unit/test_prompt.py -q   # 49452e6, env from .env.template
27 passed in 2.36s

$ (same, client.py from 8e6c60d, head tests)
FAILED tests/unit/test_prompt.py::test_ttl_zero_fetch_does_not_write_to_cache
# not run: the e2e suite against a server

After a TTL-0-only call the cache is simply cold, so the existing fallback guidance for a first call already covers it. Nothing more needed in this PR.