EPD Resolve Supplier

Versions

• v4 # Changelog ## 0.5.0 - Read `/workspace/work/code/epd-vault-index.json` (a missing or unreadable index is an empty set, never an error) and skip resolving a supplier when every one of its products is already settled by a vault document — `supplier_needs_resolving` mirrors `epd-scan-vault.short_circuits` so resolve and search agree on which suppliers still matter. New status `skipped_vault_covered`. Avoids the Ploutonion `organizations.resolve` call entirely for a fully vault-covered supplier. - `SKILL.md` updated: vault-skip prerequisite, the vault index input, and the `skipped_vault_covered` status row. ## 0.4.0 - Read `organizations.resolve` results properly. The agent answers `{candidates: [{..., identity?}]}`, which the old unwrapping did not recognise, so resolve results were dropped and suppliers fell to `no_match` unless the `identify` fallback found them. The first candidate carrying a non-empty identity now wins. - Treat an identity wire body whose `legal` and `firmographic` are both null as `no_match` rather than writing an empty `identity.json`; stamp `timestamp` when the wire body's is null. - Pass `timeout_sec=240` on both calls — resolve runs up to ~180s upstream, exactly the broker's default A2A timeout; identify up to ~120s. - Test fixtures use the real wire shapes (resolve candidates, identify body) instead of the Option-wrapped / nested-`result` shapes. ## 0.3.0 - Switched back from `organizations.findWebsite` to full Ploutonion `organizations.resolve` (with `organizations.identify` fallback when jurisdiction is known). - Writes `/workspace/work/code/supplier-identities/{supplierKey}/identity.json` per supplier. - `epd-suppliers.json` rows include `identity`, `identityPath`, and `websiteUrl` from identity. ## 0.2.1 - Per-supplier progress JSON on stdout during resolve (matches search/collect/build). - Agent must not call Ploutonion A2A skills directly — only the bundled script. ## 0.2.0 - Switched supplier resolution from `organizations.resolve` to Ploutonion's lighter `organizations.findWebsite` skill (SERP homepage discovery only, no identify enrich). Avoids the ~180s resolve budget and the broker-timeout 502 `a2a_skill_error` retries. - `epd-suppliers.json` rows now carry `discovery` (`{confidence, sources}`) instead of the full `identity` blob. Downstream `epd-search-sources` still joins on `supplierKey` + `websiteUrl` only, so its inputs are unchanged. ## 0.1.0 - Initial release: `lib/resolve_suppliers.py` reads `/workspace/work/code/epd-products.json`, deduplicates suppliers by `supplierKey`, resolves each via Ploutonion `organization-identity` (`organizations.resolve` only — no Talos), extracts `websiteUrl` from the agent response, and writes `/workspace/work/code/epd-suppliers.json`. - Status `no_match` when Ploutonion returns no usable website — downstream search uses Environdec and EPD Hub only (library-only fallback); never invents supplier URLs. 10/13/2026
• v3 # Changelog ## 0.3.0 - Switched back from `organizations.findWebsite` to full Ploutonion `organizations.resolve` (with `organizations.identify` fallback when jurisdiction is known). - Writes `/workspace/work/code/supplier-identities/{supplierKey}/identity.json` per supplier. - `epd-suppliers.json` rows include `identity`, `identityPath`, and `websiteUrl` from identity. ## 0.2.1 - Per-supplier progress JSON on stdout during resolve (matches search/collect/build). - Agent must not call Ploutonion A2A skills directly — only the bundled script. ## 0.2.0 - Switched supplier resolution from `organizations.resolve` to Ploutonion's lighter `organizations.findWebsite` skill (SERP homepage discovery only, no identify enrich). Avoids the ~180s resolve budget and the broker-timeout 502 `a2a_skill_error` retries. - `epd-suppliers.json` rows now carry `discovery` (`{confidence, sources}`) instead of the full `identity` blob. Downstream `epd-search-sources` still joins on `supplierKey` + `websiteUrl` only, so its inputs are unchanged. ## 0.1.0 - Initial release: `lib/resolve_suppliers.py` reads `/workspace/work/code/epd-products.json`, deduplicates suppliers by `supplierKey`, resolves each via Ploutonion `organization-identity` (`organizations.resolve` only — no Talos), extracts `websiteUrl` from the agent response, and writes `/workspace/work/code/epd-suppliers.json`. - Status `no_match` when Ploutonion returns no usable website — downstream search uses Environdec and EPD Hub only (library-only fallback); never invents supplier URLs. 9/22/2026
• v2 # Changelog ## 0.2.0 - Switched supplier resolution from `organizations.resolve` to Ploutonion's lighter `organizations.findWebsite` skill (SERP homepage discovery only, no identify enrich). Avoids the ~180s resolve budget and the broker-timeout 502 `a2a_skill_error` retries. - `epd-suppliers.json` rows now carry `discovery` (`{confidence, sources}`) instead of the full `identity` blob. Downstream `epd-search-sources` still joins on `supplierKey` + `websiteUrl` only, so its inputs are unchanged. ## 0.1.0 - Initial release: `lib/resolve_suppliers.py` reads `/workspace/work/code/epd-products.json`, deduplicates suppliers by `supplierKey`, resolves each via Ploutonion `organization-identity` (`organizations.resolve` only — no Talos), extracts `websiteUrl` from the agent response, and writes `/workspace/work/code/epd-suppliers.json`. - Status `no_match` when Ploutonion returns no usable website — downstream search uses Environdec and EPD Hub only (library-only fallback); never invents supplier URLs. 8/28/2026
• v1 # Changelog ## 0.1.0 - Initial release: `lib/resolve_suppliers.py` reads `/workspace/work/code/epd-products.json`, deduplicates suppliers by `supplierKey`, resolves each via Ploutonion `organization-identity` (`organizations.resolve` only — no Talos), extracts `websiteUrl` from the agent response, and writes `/workspace/work/code/epd-suppliers.json`. - Status `no_match` when Ploutonion returns no usable website — downstream search uses Environdec and EPD Hub only (library-only fallback); never invents supplier URLs. 8/27/2026