Add Tigris LangGraph checkpointer integration

[David Myriel (davidmyriel)]

Summary

Adds langgraph-checkpoint-tigris to the checkpointer integrations docs. Tigris is a globally-distributed, S3-compatible object store; the package implements a BaseCheckpointSaver (sync + async) that persists LangGraph state directly to a Tigris bucket, plus a zero-copy bucket fork that branches an agent's entire checkpoint history in O(1).

Changes:

• Add a row to the checkpointer integrations table.
• Add a Tigris provider page (providers/tigris.mdx) covering the sync/async saver and the zero-copy fork, with the Single/Multi-region bucket consistency requirement called out.
• Add a Tigris card to the all-providers list.

The package is maintained in tigrisdata/tigris-langgraph and follows the same third-party pattern as the Redis, CockroachDB, and Aerospike entries (package + provider page, not vendored into langgraph).

Why Tigris

Unlike the existing database-backed checkpointers, Tigris is pure object storage (no database to operate, no egress fees) and supports zero-copy forking: fork() branches every thread and checkpoint instantly with writes isolated from the source — useful for parallel evals, what-if exploration, and isolated experiments against a real agent's accumulated state.

Validation

Against a live Tigris bucket, the package passes:

• the LangGraph checkpointer conformance suite (langgraph-checkpoint-conformance), sync and async;
• delta-channel reconstruction tests (sync + async) — verifying the direct by-checkpoint_id get_tuple lookup and full ancestor-chain copy_thread that delta channels depend on;
• end-to-end resume + time travel, zero-copy fork isolation, copy_thread, and list ordering/limits.

Test plan

• make lint_prose / make lint_md pass on the changed files
• Provider page renders (make dev) and the table link resolves to #langgraph-checkpointer
• Links to PyPI and the source repo resolve

[github-actions]

Thanks for opening a docs PR, David Myriel (@davidmyriel)! When it's ready for review, please add the relevant reviewers:

• @mdrxy (Python integrations)

[David Myriel (davidmyriel)]

Hi Mason Daugherty (@mdrxy), thank you for looking into this one. I was advised by a team members via forums to PR here.
At Tigris we have storage-based checkpointing thats very efficient and used by a lot of agent builders. I thought it would be useful here as well. We published a package to support this the other day and will continue to add advanced features for agents.