bd query — “query issues using a simple query language that supports compound filters, boolean operators, and date-relative expressions”, the escape hatch when beads-list’s flags stop composing. bd search <text> is the plain text search beside it.
Three duplicate-detection commands share the page because they are one idea: bd find-duplicates finds semantically similar beads “using text analysis or AI”, bd duplicates finds and optionally merges them, and bd duplicate <a> <b> marks one as a duplicate of another — which is a knowledge-graph link, not a deletion (beads-supersede).
Duplicate detection earns its place in a store rather than a framework: a work graph that several agents append to over months accumulates near-identical beads, and nothing upstream prevents it. This is the tracker maintaining its own signal-to-noise.
bd sql — raw SQL against the underlying database, “useful for debugging, maintenance, and working around bugs in higher-level commands” — is the unrestricted version of this surface, and is documented with beads-dolt.
Maps to no canonical SDLC stage.
See Also
- beads-list — the flag-based read.
- beads-supersede — where a confirmed duplicate link lands.
- beads-dolt — the raw SQL surface underneath.