Reported issues for Plant Genomics

Pod holds 13 of 13 GitHub reports that passed its relevance review. This can include external user reports, maintainer-confirmed bugs, and concrete feature gaps. Treat them as evidence to inspect, not a count of distinct defects.

Back to Plant Genomics.

Most discussed

Normalisation requests: six values this server passes through from upstream unchanged

Written from the ARF family dossier in examples/arf_family/, re-run at server commit 967bc36 (release 1.21.0): 400 MCP calls over 48 genes x 16 tools, the family itself from a 742-call enumeration through the same tools. Each section below is one row of gaps.jsonl, quoted as logged; Check names the captured response…

Read the thread · 2026-09-22 · closed · 2 comments

`locus_literature` promises a journal and returns null in all 160 hits

Read the thread · 2026-09-22 · closed · 1 comment

Family, entry and gene-tree accessions come back from tools that nothing can take them into

Read the thread · 2026-09-22 · closed · 1 comment

Both ortholog tools cap at 100 rows with no target-organism filter, so rice and wheat never come back

Written from the ARF family dossier run in examples/arf_family/, server version 1.21.0, 2026-09-21 (29 genes x 16 tools, 248 MCP calls; the family itself from a 516-call enumeration through the same tools). Each section below is one row of gaps.jsonl, quoted as logged; Check names the captured response to read it back…

Read the thread · 2026-09-22 · closed · 1 comment

`gene_report`: payloads carried twice, per-step timings always null, two names for one gene, GO bullets repeated

Read the thread · 2026-09-22 · closed · 1 comment

Every wheat IWGSC locus is a 404 in STRING, though the same loci resolve to UniProt

Written from the ARF family dossier in examples/arf_family/, re-run at server commit 052e4ec (release 1.21.0): 64 MCP calls over 114 genes x 16 tools, the family itself from a 1163-call enumeration through the same tools. Each section below is one row of gaps.jsonl, quoted as logged; Check names the captured…

Read the thread · 2026-09-23 · closed · 0 comments

`gene_report` answers ok with an empty result when its Ensembl step fails, where the description promises a degraded dossier

Read the thread · 2026-09-23 · closed · 0 comments

`batch_locus_call` fans out 8 wide, and OrthoDB refuses the excess as a rate limit

Read the thread · 2026-09-23 · closed · 0 comments

Most recent

`string_interactions.accession` is a STRING id, not the UniProt accession the other tools mean by that name

Read the thread · 2026-09-22 · closed · 0 comments

`locus_go_annotations` drops a row without a flag and ships a dedup the payload does not label

Read the thread · 2026-09-22 · closed · 0 comments

81% of responses report no upstream release, and three report one under a key nothing reads

Read the thread · 2026-09-22 · closed · 0 comments

fuzz: 1 crash(es) found 2026-09-20

Automated fuzzing (hypothesis) of pure entry-point helpers. Note: this session's sandbox denied all filesystem writes (Write tool, shell redirection, mkdir, even Python's open() were blocked), so tests ran as inline uv run python -c snippets rather than files under fuzz_scratch/ as normally required — no files could be committed to that directory this run.

Crash 1: `organisms.resolve()` — unhandled AttributeError/TypeError on non-str/int input

organisms.resolve()…

Read the thread · 2026-09-20 · closed · external user · 0 comments

fuzz: 2 crash(es) found 2026-09-14

Hypothesis property-based fuzzing of pure, network-free entry points in plant_genomics_mcp (validators, organisms, batch, cache, synthesis reconciliation helpers). Ran with --hypothesis-seed=0, no fixed test files were persisted this run (sandbox write permissions were unavailable), tests were executed inline via uv run python with the same strategies described below.

Both crashes are unhandled Python exceptions (AttributeError / TypeError) rather than a deliberate…

Read the thread · 2026-09-14 · closed · external user · 0 comments

The remaining reports are on the project's issue tracker.

Reported issues for Plant Genomics

Most discussed

Normalisation requests: six values this server passes through from upstream unchanged

locus_literature promises a journal and returns null in all 160 hits

Family, entry and gene-tree accessions come back from tools that nothing can take them into

Both ortholog tools cap at 100 rows with no target-organism filter, so rice and wheat never come back

gene_report: payloads carried twice, per-step timings always null, two names for one gene, GO bullets repeated

Every wheat IWGSC locus is a 404 in STRING, though the same loci resolve to UniProt

gene_report answers ok with an empty result when its Ensembl step fails, where the description promises a degraded dossier

batch_locus_call fans out 8 wide, and OrthoDB refuses the excess as a rate limit

Most recent

string_interactions.accession is a STRING id, not the UniProt accession the other tools mean by that name

locus_go_annotations drops a row without a flag and ships a dedup the payload does not label

81% of responses report no upstream release, and three report one under a key nothing reads

fuzz: 1 crash(es) found 2026-09-20

Crash 1: organisms.resolve() — unhandled AttributeError/TypeError on non-str/int input

fuzz: 2 crash(es) found 2026-09-14

`locus_literature` promises a journal and returns null in all 160 hits

`gene_report`: payloads carried twice, per-step timings always null, two names for one gene, GO bullets repeated

`gene_report` answers ok with an empty result when its Ensembl step fails, where the description promises a degraded dossier

`batch_locus_call` fans out 8 wide, and OrthoDB refuses the excess as a rate limit

`string_interactions.accession` is a STRING id, not the UniProt accession the other tools mean by that name

`locus_go_annotations` drops a row without a flag and ships a dedup the payload does not label

Crash 1: `organisms.resolve()` — unhandled AttributeError/TypeError on non-str/int input