Entrypoints and Examples¶
The installed bijux-pollenomics console script is the supported runtime
entrypoint. Commands fall into inspection, validation, collection, evidence
refresh, and publication classes. Choose the narrowest class that answers the
question because only some classes rewrite governed state.
Discover The Interface¶
The executable resolves from the active environment. Record the reported runtime version before creating evidence or publication artifacts.
Verification Entry Points¶
bijux-pollenomics source-support --json
bijux-pollenomics adna-species
bijux-pollenomics adna-species-review --species ovis_aries --json
bijux-pollenomics adna-runtime-manifest --species ovis_aries --json
These commands inspect registered source or species state. They do not perform a collection or publication refresh. Capture the command, runtime version, explicit roots when applicable, and emitted identity so another reader can repeat the same verification against the same state.
Validate A Collection Ledger¶
Validation checks the existing summary and is the appropriate first response to a summary-contract question.
Collection And Publication Examples¶
Use explicit isolated roots when learning a state-changing command. The following evaluation keeps collected evidence and publications outside the governed repository trees:
bijux-pollenomics collect-data all \
--version v66 \
--output-root artifacts/operator-evaluation/data
bijux-pollenomics publish-reports \
--aadr-root artifacts/operator-evaluation/data/aadr \
--version v66 \
--context-root artifacts/operator-evaluation/data \
--output-root artifacts/operator-evaluation/reports
collect-data writes source-family trees and the collection ledger under the
selected data root. publish-reports consumes that explicit state and writes
world, regional, country, review, and caveat products under the selected report
root. Publication does not recollect a source implicitly.
Read A Publication Packet¶
For the checked-in repository state, begin with the publication summary and a named bundle rather than opening HTML first:
jq '{schema_version, version, countries,
geography_bundle_count: (.geography_bundles | length),
artifact_count: (.artifacts | length)}' \
docs/report/published_reports_summary.json
jq '{scope_key, scope_kind, countries, total_unique_samples,
animal_points: .animal_atlas.total_locality_points}' \
docs/report/world/world_bundle.json
jq '{overall_ok, reference_grade_claim_allowed,
reference_grade_support_ready}' \
docs/report/animal_publication_release_gate.json
The summary identifies the publication family, the bundle identifies one scope and its populations, and the release gate constrains the language that may be used for that state. None replaces member-level traceability or the governing evidence record. A coherent bundle with a false release claim flag is an accountable product with restricted claim language, not a failed JSON read.
flowchart LR
Command["state-changing command"] --> Inputs["explicit input roots"]
Inputs --> Isolated["isolated evaluation outputs"]
Isolated --> Review["manifest, evidence, semantic, and membership review"]
Review --> Promote{"intended governed replacement?"}
Promote -->|no| Retain["diagnostic evaluation only"]
Promote -->|yes| Governed["use the repository-owned regeneration workflow"]
Copying selected evaluation files into governed roots is not promotion. A governed replacement must use the owning workflow and review the complete causal diff.
Compare An Evaluation With Governed State¶
Before deciding whether an isolated run should become repository state, compare it with the current governed baseline:
| Comparison | Evidence to inspect |
|---|---|
| source identity | release, retrieval context, member inventory, content digests, and access posture |
| governed objects | added, removed, retained, merged, split, and superseded stable IDs |
| meaning | units, nulls, roles, precision, time basis, geometry basis, and relations |
| review | conflicts, qualifications, refusals, and recovery conditions |
| populations | captured, eligible, reviewed, admitted, excluded, and published denominators |
| products | manifest membership, caveats, structured members, and rendered descendants |
If the comparison cannot attribute a product change to source, evidence, decision, boundary, or product-rule input, keep the evaluation isolated. A visually plausible map is not enough to replace governed state.
Atlas And Country Surfaces¶
bijux-pollenomics report-country Sweden \
--aadr-root artifacts/operator-evaluation/data/aadr \
--version v66 \
--context-root artifacts/operator-evaluation/data \
--output-root artifacts/operator-evaluation/reports
bijux-pollenomics report-multi-country-map Sweden Norway Finland Denmark \
--aadr-root artifacts/operator-evaluation/data/aadr \
--version v66 \
--context-root artifacts/operator-evaluation/data \
--output-root artifacts/operator-evaluation/reports
report-country writes one country bundle. report-multi-country-map writes a
shared map for the named countries. They are narrower publication operations,
not shortcuts around the same evidence requirements.
Mutation Boundaries¶
| Command family | Reads | Writes |
|---|---|---|
source-support, adna-* inspection |
current governed state | standard output |
validate-collection-summary |
one summary file | standard output and exit status |
collect-data |
external sources and collector configuration | data/ |
refresh-data-contract-surfaces |
current data tree | collection and contract summaries |
publish-reports, report-* |
governed data and geography configuration | docs/report/ |
Inspect the resulting diff after every state-changing command. A successful exit establishes command completion; the evidence and publication diffs still require review.
The table names the repository-owned default roots. Explicit paths override those defaults, as in the isolated examples above. Before execution, resolve every relative path against the current working directory and record the full input and output identities.