# v0.99.0: Verified capabilities and product truth > **Status:** Planned > **Scope:** Medium > **User promise:** *"I can determine whether my query will work, how pg_trickle will maintain it, and which limits apply."* > **Blocked by:** [v0.98.x](v0.98.x.md) > **Source:** [September 2026 assessment](../plans/pg-trickle-assessment-and-pre-1.0-roadmap.md) ## Theme Make the implementation, public claims, and release evidence agree. This release adds no SQL coverage or integrations. It establishes the baseline used to judge every later pre-1.0 change. ## Work - Generate one machine-readable capability and strategy manifest from runnable admission examples and stable reason identifiers. - Use the manifest to reconcile the README, support matrix, SQL reference, generated catalogs, configuration docs, and integration docs. - Replace nominal Graph V1 checks with PostgreSQL tests for real graphs, authorization, strategy policy, rollback, and coordinator writes. - Record release evidence against a source commit, artifact digest, PostgreSQL version, suite version, workload, result, and retained logs. - Add regression coverage for WAL receipt durability, graph strategy policy, graph boundary semantics, source-lock behavior, and backend-local cleanup. - Mark an unproven stable capability unavailable or experimental until its conformance tests pass. ## Exit criteria - [ ] Every public support claim maps to a runnable example and its expected strategy or rejection. - [ ] No P0 correctness investigation remains open for an enabled stable capability. - [ ] Graph policy, authorization, and rollback tests execute the public SQL API against PostgreSQL. - [ ] Documentation and catalog checks pass without broad exemptions for active unsupported interfaces. - [ ] The evidence manifest distinguishes executed, skipped, unavailable, and historical results. ## Scope control Do not add SQL coverage or integrations. Fix urgent correctness and security defects in the earliest safe release instead of reserving them for this one.