Found in the adversarial review of #120.
resources.md pipeline diagram still shows every Mech feeding kg-microbe, which contradicts the rewritten hub claim
Where: resources.md:136
The PR now says 'Only MediaIngredientMech feeds kg-microbe today', that CultureMech and CommunityMech have no published release, and that the other six release no data into the graph. It also rewrote the /mechs/ Related-resources line away from 'the Mechs publish into'. But the 'Pipeline Overview' diagram in resources.md still draws a single flow: CultureMech and MIM, then CommunityMech and the other domain Mechs, then KG-Microbe Knowledge Graph. That diagram sits about 15 lines above text the PR edited, and the PR left it unchanged.
Failure scenario. A reader on /resources/#-ai-curation-tools is told that CultureMech, CommunityMech and the other Mechs flow into KG-Microbe. The same site's /mechs/ hub panel says only MIM does, so the two pages give contradictory accounts of the data flow this PR set out to correct.
Evidence. resources.md:136-148 (unchanged vs main:136-148): 'CultureMech ↔ MediaIngredientMech ... ↓ CommunityMech + other domain-specific Mechs ... ↓ KG-Microbe Knowledge Graph'. _fleet/fleet_fragment.html:785: 'Only MediaIngredientMech feeds kg-microbe today ... The other six share the kg-microbe namespace but release no data into the graph.' I verified the hub claim independently. In a shallow clone of Knowledge-Graph-Hub/kg-microbe master (b8161c4), grep for every Mech name finds data inputs only from MIM (mappings/ingredient_mappings.sssom.tsv and culturebotai_reviewed_ingredients.tsv, synced from MIM's UNIFIED_INGREDIENT_MAPPING.tsv). gh release list returns no releases for CultureMech or CommunityMech.
Verifier (confirmed, low). The unchanged resources.md Pipeline Overview (lines 136-148) routes CommunityMech and the other domain-specific Mechs into the KG-Microbe Knowledge Graph. That contradicts the PR's rewritten /mechs/ hub text: only MIM's SSSOM feeds kg-microbe, CultureMech and CommunityMech have no release, and the rest release no data into the graph. CultureMech content does reach kg-microbe, but only indirectly, as provenance inside MIM's mappings. So the redraw should show CultureMech → MIM → kg-microbe, with the other Mechs consuming from kg-microbe. The diagram predates the PR (d4e424d). This is an unfinished propagation of the PR's own correction, not something the PR introduced.
Suggested fix. Redraw the diagram so MIM alone feeds KG-Microbe, and show kg-microbe supplying data back to the Mechs (or drop the KG-Microbe step), to match the hub panel.
Found in the adversarial review of #120.
resources.md pipeline diagram still shows every Mech feeding kg-microbe, which contradicts the rewritten hub claim
Where:
resources.md:136The PR now says 'Only MediaIngredientMech feeds kg-microbe today', that CultureMech and CommunityMech have no published release, and that the other six release no data into the graph. It also rewrote the /mechs/ Related-resources line away from 'the Mechs publish into'. But the 'Pipeline Overview' diagram in resources.md still draws a single flow: CultureMech and MIM, then CommunityMech and the other domain Mechs, then KG-Microbe Knowledge Graph. That diagram sits about 15 lines above text the PR edited, and the PR left it unchanged.
Failure scenario. A reader on /resources/#-ai-curation-tools is told that CultureMech, CommunityMech and the other Mechs flow into KG-Microbe. The same site's /mechs/ hub panel says only MIM does, so the two pages give contradictory accounts of the data flow this PR set out to correct.
Evidence. resources.md:136-148 (unchanged vs main:136-148): 'CultureMech ↔ MediaIngredientMech ... ↓ CommunityMech + other domain-specific Mechs ... ↓ KG-Microbe Knowledge Graph'. _fleet/fleet_fragment.html:785: 'Only MediaIngredientMech feeds kg-microbe today ... The other six share the kg-microbe namespace but release no data into the graph.' I verified the hub claim independently. In a shallow clone of Knowledge-Graph-Hub/kg-microbe master (b8161c4), grep for every Mech name finds data inputs only from MIM (mappings/ingredient_mappings.sssom.tsv and culturebotai_reviewed_ingredients.tsv, synced from MIM's UNIFIED_INGREDIENT_MAPPING.tsv).
gh release listreturns no releases for CultureMech or CommunityMech.Verifier (confirmed, low). The unchanged resources.md Pipeline Overview (lines 136-148) routes CommunityMech and the other domain-specific Mechs into the KG-Microbe Knowledge Graph. That contradicts the PR's rewritten /mechs/ hub text: only MIM's SSSOM feeds kg-microbe, CultureMech and CommunityMech have no release, and the rest release no data into the graph. CultureMech content does reach kg-microbe, but only indirectly, as provenance inside MIM's mappings. So the redraw should show CultureMech → MIM → kg-microbe, with the other Mechs consuming from kg-microbe. The diagram predates the PR (d4e424d). This is an unfinished propagation of the PR's own correction, not something the PR introduced.
Suggested fix. Redraw the diagram so MIM alone feeds KG-Microbe, and show kg-microbe supplying data back to the Mechs (or drop the KG-Microbe step), to match the hub panel.