A token scoped to a single script or flow path (e.g.
`scripts:read:f/allowed/*`) could call `GET .../scripts/list_search` (or
`/list`) and receive `path` + full `content` for every script the
underlying user could see — likewise `flows/list_search` leaked the full
flow `value`. Route-level scope checks only validate `domain:action`, and
the listing handlers did no per-row scope filtering, leaking out-of-scope
source/definitions to narrowly-scoped tokens.
Apply `build_scope_path_predicate` (added in #9302 for resources/variables)
to `list_search_scripts`, `list_scripts`, `list_search_flows`, and
`list_flows`, mirroring the resources/variables fix exactly. Unscoped
tokens and tokens whose only scopes are `if_jobs:filter_tags:*` are
unaffected.
Adds integration regression tests (scripts + flows) covering: path-scoped
token sees only in-scope paths, broad `*:read` token still sees all
RLS-visible items, tag-filter-only and unscoped tokens unchanged.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Ruben Fiszel