accounted

admin/accounted

Fork 0

Commit Graph

Author	SHA1	Message	Date
Jakob WennbergandClaude Opus 4.7	6629964780	fix(mcp): unblock query_journal text search across line + entry descriptions (#605 ) * fix(mcp): unblock query_journal text search across line+entry descriptions `gnubok_query_journal` issued a single PostgREST `.or()` filter mixing a base-table column (line_description on journal_entry_lines) with a column on the embedded inner-joined resource (journal_entries.description). PostgREST's flat comma OR syntax cannot span base and embedded resources and returned 400 "failed to parse logic tree" on every free-text search, so the agent could not look up history (e.g. "har vi bokfört Google innan?") via the MCP tool. Refactor: pull the common filter chain into a `buildBaseQuery()` helper and issue two parallel `.ilike()` queries — one on `line_description`, one on `journal_entries.description` — then merge by line id, re-sort, and slice to `limit`. Same pattern as `lib/invoices/duplicate-payment-candidates.ts`, with the same rationale (LIKE-DSL injection risk + cross-embed unsupported). Drops the obsolete comma-stripping pre-processing on the search term (only needed because the value was being injected into PostgREST's OR DSL); LIKE wildcard escaping (`%` and `_`) is preserved. Adds four tests covering: merge across both legs, dedup of cross-leg overlap, correct columns + escaped pattern sent to `.ilike`, and LIKE-wildcard escape end-to-end. Existing amount-filter / accounts-cap / truncation tests unchanged. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(mcp): query_journal PR review — honest truncation, wider leg, error sanitisation Address PR #605 review (Greptile + compliance swarm). - truncated math: previously dbMatched = byLine.count + byEntry.count inflated the counter every time a row matched both legs (e.g. an entry whose line description AND header description both contain "Google"). total_lines was off by up to 2× and truncated was forced true even when every distinct match was already returned, sending the agent into unnecessary narrow-and-retry loops. dbMatched is now merged.size — the honest distinct-row count among what we fetched. A new legCapHit signal drives truncated when either leg's fetch window filled, so we never lie the other direction either. - per-leg fetch widened to 2 × limit. Previously each leg was capped at `limit` independently, so when one column was much more selective than the other (e.g. 150 line matches vs 5 entry matches), the merge could drop globally-ranked rows from the chronologically newer tail. The final slice still caps at `limit`; the wider per-leg window just gives the merge a better tail to choose from. - text input length cap (200 chars). Defence-in-depth against pathological inputs even though .ilike() parameterises the value (compliance A.8.28). - error sanitisation in the text-search path. Raw PostgREST messages can surface internal schema details (table names, constraint names) to the caller. We now log details server-side via the existing module logger and throw a generic message (compliance V16). - comma-stripping rationale documented inline so the next reviewer doesn't re-add it. The previous defence was needed because the value was injected into PostgREST's OR DSL where `,` is the separator; the parameterised .ilike() path treats `,` as a literal and stripping it would mangle real-comma searches (compliance V1.2.5). Adds 4 regression tests: overlap-doesn't-falsely-truncate, leg-cap-hit flags truncated, text>200 rejected, raw schema name never leaked. All 14 query_journal tests pass; 266/266 MCP suite green; build clean. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(mcp): query_journal PR review round 2 — schema guards, leg cap, audit log Address compliance swarm findings on commit 3dfab46c. - text input: add `maxLength: 200` to the JSON Schema so the limit is enforced at the boundary by the MCP runtime, not only inside execute(). Inline guard kept as belt-and-suspenders (ASVS V2.2, ISO 27001 A.8.28). - limit input: add `minimum: 1, maximum: 500` to the JSON Schema (matches the existing Math.min cap inside execute()) so the bound is visible to agents and enforced declaratively. - legLimit hard cap: `Math.min(limit * 2, 500)`. Previously legLimit scaled with limit unbounded, so a maxed-out limit triggered 2×1000-row fetches. Cap is independent of the caller-supplied limit (ASVS V2.3). - audit log: include `userId` alongside `companyId` in both log.warn payloads so failed-query events can be correlated to the actor for incident investigation (ASVS V16, SOC 2 CC7.2). - new test: assert both parallel ilike legs in the text-search path issue `.eq('journal_entries.company_id', companyId)` — defence-in-depth against a future refactor accidentally dropping tenant scoping from one leg (SOC 2 CC6.1). Not addressed (with rationale): - ASVS V1.2.5 / SOC 2 CC7.2 "log.warn body contains raw PostgREST error text". The swedish-accounting-compliance reviewer explicitly cleared this on the same commit ("the schema-leak is plugged in both paths"). The thrown error is generic; raw detail lives only in internal structured logs (Sentry, etc.) where it is needed for incident triage. Same pattern is used by 20+ other tools in this file; scrubbing only this one is inconsistent. Defer to a project-wide logger sanitiser. - Swedish reviewer's "reversed mixed with posted in status=all": false positive — `status` IS surfaced per line in the response shape (see the lines.map projection in server.ts). - Swedish reviewer's "debit XOR credit not enforced": insert-time concern, not a read-tool concern. Out of scope. Tests: 15 query_journal cases (4 new this round); 267 MCP suite; build clean. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-05-29 14:21:46 +02:00
Jakob WennbergandClaude Opus 4.7	4131db2894	chore: MCP intent-tools, BankID enrichment table, multi-tenant fixes (#402 ) * chore: MCP intent-tools, BankID enrichment table, multi-tenant fixes MCP server gains six intent-shaped tools that collapse multi-call agent flows into one: vat_close_check, query_journal, auto_match_period, create_supplier_invoice_from_inbox, audit_package, year_end_readiness. Tools wired into TOOL_SCOPE_MAP and OPERATION_RISK_TIERS as appropriate (create_supplier_invoice_from_inbox at medium tier — reversible until approve, but stages a leverantörsskuld). BankID enrichment now persists to a dedicated bankid_enrichment table keyed by user_id. extension_data has been company-scoped (NOT NULL company_id) since the multi-tenant refactor, so every BankID signup has silently been failing the enrichment upsert. Select-company picker reads from the new table. delete_last_voucher (BFNAR 2013:2) needs to clear document_attachments.journal_entry_id before deleting the entry, but the new document immutability trigger blocks that UPDATE. Added the same gnubok.allow_delete transaction-scoped bypass pattern used by the journal-entry/line/retention triggers. pg-real tests cover the happy path, the unauthorized direct UPDATE, and the swap-to-different-entry attempt under the bypass flag. fiscal_periods.no_overlapping_fiscal_periods exclusion was scoped to user_id from before multi-tenant — rebound to company_id so the same user can have overlapping fiscal years across companies they own/are member of. Also adds scripts/seed-demo-account.ts for end-to-end demo seeding (two companies, full FY2025, active FY2026 with mixed state). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(pr-402): address review feedback Migrations - Drop 20260506140000_document_journal_entry_immutability_delete_bypass.sql: redundant with 20260506140000_document_journal_entry_immutability_bypass.sql that landed on main while this branch was open. Both share the same gnubok.allow_delete pattern; main's version is what the DB actually has. - Rename 20260506150000_bankid_enrichment_table.sql → 20260506160000_bankid_enrichment_table.sql to clear the timestamp clash with 20260506150000_protect_document_journal_link.sql on main (Supabase branch preview was failing on schema_migrations PK collision). Tests - Drop the swap-under-flag test from delete-last-voucher.pg.test.ts: main's bypass returns NEW unconditionally when gnubok.allow_delete='true', so the swap is permitted. Drop the duplicate happy-path test (already covered by 'clears journal_entry_id on attached documents and deletes the voucher'). Keep the unauthorized-direct-UPDATE test. - Add bankid-enrichment.pg.test.ts covering the SELECT RLS policy: user reads own row, cannot read another user's row, INSERT denied for authenticated. gnubok_query_journal - amount_min/amount_max is applied post-fetch (PostgREST can't OR abs(debit) and abs(credit) cleanly), but PostgREST's count is computed pre-filter. Reporting that as total_lines mislead agents into paginating a tail that was already filtered out. When the amount filter is applied, anchor total_lines and truncated to the filtered set and surface db_matched_pre_amount_filter + amount_filter_applied_post_fetch separately. - Escape `_` in the free-text LIKE filter so a search for "2_441" doesn't match "2X441". VAT close check - Reverse-charge blocker no longer fires on ruta 30 (seller-side domestic omvänd skattskyldighet) — the seller books no VAT, the buyer does, so missing ruta 48 is expected. Now scoped to ruta 31/32 (EU acquisition) where the buyer must book both calculated output (2615) and matching ingående moms (2645). - High-value receipt threshold no longer reads journal_entries.total_amount (column doesn't exist; check silently never fired). Sums debits across the entry's lines, which equals the gross for ordinary purchase entries — comparing a gross figure against the BFL/ML 4 000 SEK threshold per ML 17 kap 26–28 §. seed-demo-account.ts - Require an explicit email argument; refuse to run with the previously hardcoded fallback that would silently target a real user. Ensure email is non-undefined for downstream typing. - Type the supabase fiscal_periods insert result locally so tsc no longer reports 'fp implicitly any' from the loose untyped client. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(test): adjust fiscal-period-start-day pg test for per-company overlap The pg-real failure on PR #402 was a latent bug surfaced by this branch's fiscal_periods exclusion constraint flip from user_id to company_id (migration 20260506140100). The test was inserting periods that overlapped seedCompany's default 2026-01-01..2026-12-31 period; the previous constraint slipped past it because the test's INSERT didn't set user_id (NULL escapes the WITH = match), so two same-company overlapping periods silently coexisted. Now that the constraint correctly fires per company, pick years that don't overlap with the seeded 2026 period. The trigger's behavior under test (allow mid-month start when no earlier period exists, allow back-dated SIE imports, reject mid-month start when an earlier period exists) is unchanged. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(vat-close-check): correct reverse-charge/import blocker rutor Rutor 30/31/32 are the buyer's calculated utgående moms on reverse- charge purchases (domestic byggtjänster/electronics → 2614 → ruta 30; EU goods → 2624 → ruta 31; EU services → 2634 → ruta 32). The buyer must also book matching ingående moms (2647 inhemskt / 2645 utlandet → ruta 48). The previous fix removed ruta 30 on the basis that it was seller-side; that's incorrect — domestic-RC sellers book no VAT at all (they report only beskattningsunderlag on ruta 41), so 2614 only sees buyer-side entries. Restore ruta 30. Also extend the check to import rutor 60/61/62 (non-EU import VAT declared via momsdeklaration since 2015 — 2615/2625/2635). Same mechanic: importer books output VAT on these rutor and deducts the input side via ruta 48. SaaS-from-AWS / OpenAI / Vercel companies hit this path; without including 60/61/62 the blocker would silently miss their misbookings. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(mcp): expose ruta 60/61/62 (import VAT) on the local VatReportResult The vat-close-check fix referenced vatReport.rutor.ruta60/61/62 but the MCP server's local VatReportResult type only carries ruta 05-49. Build broke on tsc. Extend the MCP server's slim VAT report to also project import VAT — 2615 → ruta 60 (25%), 2625 → ruta 61 (12%), 2635 → ruta 62 (6%) — and fold those into ruta 49 (att betala/återfå). Mirrors the BAS-to-Ruta mapping in lib/reports/vat-declaration.ts. Output schema and required list updated accordingly. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-05-06 16:41:36 +02:00

Author

SHA1

Message

Date

Jakob WennbergandClaude Opus 4.7

6629964780

fix(mcp): unblock query_journal text search across line + entry descriptions (#605 )

* fix(mcp): unblock query_journal text search across line+entry descriptions

`gnubok_query_journal` issued a single PostgREST `.or()` filter mixing a
base-table column (line_description on journal_entry_lines) with a column
on the embedded inner-joined resource (journal_entries.description).
PostgREST's flat comma OR syntax cannot span base and embedded resources
and returned 400 "failed to parse logic tree" on every free-text search,
so the agent could not look up history (e.g. "har vi bokfört Google
innan?") via the MCP tool.

Refactor: pull the common filter chain into a `buildBaseQuery()` helper
and issue two parallel `.ilike()` queries — one on `line_description`,
one on `journal_entries.description` — then merge by line id, re-sort,
and slice to `limit`. Same pattern as
`lib/invoices/duplicate-payment-candidates.ts`, with the same rationale
(LIKE-DSL injection risk + cross-embed unsupported).

Drops the obsolete comma-stripping pre-processing on the search term
(only needed because the value was being injected into PostgREST's OR
DSL); LIKE wildcard escaping (`%` and `_`) is preserved.

Adds four tests covering: merge across both legs, dedup of cross-leg
overlap, correct columns + escaped pattern sent to `.ilike`, and
LIKE-wildcard escape end-to-end. Existing amount-filter / accounts-cap /
truncation tests unchanged.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* fix(mcp): query_journal PR review — honest truncation, wider leg, error sanitisation

Address PR #605 review (Greptile + compliance swarm).

- truncated math: previously dbMatched = byLine.count + byEntry.count
  inflated the counter every time a row matched both legs (e.g. an entry
  whose line description AND header description both contain "Google").
  total_lines was off by up to 2× and truncated was forced true even when
  every distinct match was already returned, sending the agent into
  unnecessary narrow-and-retry loops. dbMatched is now merged.size — the
  honest distinct-row count among what we fetched. A new legCapHit signal
  drives truncated when either leg's fetch window filled, so we never lie
  the other direction either.

- per-leg fetch widened to 2 × limit. Previously each leg was capped at
  `limit` independently, so when one column was much more selective than
  the other (e.g. 150 line matches vs 5 entry matches), the merge could
  drop globally-ranked rows from the chronologically newer tail. The
  final slice still caps at `limit`; the wider per-leg window just gives
  the merge a better tail to choose from.

- text input length cap (200 chars). Defence-in-depth against pathological
  inputs even though .ilike() parameterises the value (compliance A.8.28).

- error sanitisation in the text-search path. Raw PostgREST messages can
  surface internal schema details (table names, constraint names) to the
  caller. We now log details server-side via the existing module logger
  and throw a generic message (compliance V16).

- comma-stripping rationale documented inline so the next reviewer
  doesn't re-add it. The previous defence was needed because the value
  was injected into PostgREST's OR DSL where `,` is the separator; the
  parameterised .ilike() path treats `,` as a literal and stripping it
  would mangle real-comma searches (compliance V1.2.5).

Adds 4 regression tests: overlap-doesn't-falsely-truncate, leg-cap-hit
flags truncated, text>200 rejected, raw schema name never leaked. All
14 query_journal tests pass; 266/266 MCP suite green; build clean.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* fix(mcp): query_journal PR review round 2 — schema guards, leg cap, audit log

Address compliance swarm findings on commit 3dfab46c.

- text input: add `maxLength: 200` to the JSON Schema so the limit is
  enforced at the boundary by the MCP runtime, not only inside execute().
  Inline guard kept as belt-and-suspenders (ASVS V2.2, ISO 27001 A.8.28).
- limit input: add `minimum: 1, maximum: 500` to the JSON Schema (matches
  the existing Math.min cap inside execute()) so the bound is visible to
  agents and enforced declaratively.
- legLimit hard cap: `Math.min(limit * 2, 500)`. Previously legLimit
  scaled with limit unbounded, so a maxed-out limit triggered 2×1000-row
  fetches. Cap is independent of the caller-supplied limit (ASVS V2.3).
- audit log: include `userId` alongside `companyId` in both log.warn
  payloads so failed-query events can be correlated to the actor for
  incident investigation (ASVS V16, SOC 2 CC7.2).
- new test: assert both parallel ilike legs in the text-search path issue
  `.eq('journal_entries.company_id', companyId)` — defence-in-depth
  against a future refactor accidentally dropping tenant scoping from
  one leg (SOC 2 CC6.1).

Not addressed (with rationale):

- ASVS V1.2.5 / SOC 2 CC7.2 "log.warn body contains raw PostgREST error
  text". The swedish-accounting-compliance reviewer explicitly cleared
  this on the same commit ("the schema-leak is plugged in both paths").
  The thrown error is generic; raw detail lives only in internal
  structured logs (Sentry, etc.) where it is needed for incident triage.
  Same pattern is used by 20+ other tools in this file; scrubbing only
  this one is inconsistent. Defer to a project-wide logger sanitiser.

- Swedish reviewer's "reversed mixed with posted in status=all": false
  positive — `status` IS surfaced per line in the response shape (see
  the lines.map projection in server.ts).

- Swedish reviewer's "debit XOR credit not enforced": insert-time
  concern, not a read-tool concern. Out of scope.

Tests: 15 query_journal cases (4 new this round); 267 MCP suite; build
clean.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-05-29 14:21:46 +02:00

Jakob WennbergandClaude Opus 4.7

4131db2894

chore: MCP intent-tools, BankID enrichment table, multi-tenant fixes (#402 )

* chore: MCP intent-tools, BankID enrichment table, multi-tenant fixes

MCP server gains six intent-shaped tools that collapse multi-call
agent flows into one: vat_close_check, query_journal, auto_match_period,
create_supplier_invoice_from_inbox, audit_package, year_end_readiness.
Tools wired into TOOL_SCOPE_MAP and OPERATION_RISK_TIERS as appropriate
(create_supplier_invoice_from_inbox at medium tier — reversible until
approve, but stages a leverantörsskuld).

BankID enrichment now persists to a dedicated bankid_enrichment table
keyed by user_id. extension_data has been company-scoped (NOT NULL
company_id) since the multi-tenant refactor, so every BankID signup has
silently been failing the enrichment upsert. Select-company picker reads
from the new table.

delete_last_voucher (BFNAR 2013:2) needs to clear
document_attachments.journal_entry_id before deleting the entry, but the
new document immutability trigger blocks that UPDATE. Added the same
gnubok.allow_delete transaction-scoped bypass pattern used by the
journal-entry/line/retention triggers. pg-real tests cover the happy
path, the unauthorized direct UPDATE, and the swap-to-different-entry
attempt under the bypass flag.

fiscal_periods.no_overlapping_fiscal_periods exclusion was scoped to
user_id from before multi-tenant — rebound to company_id so the same
user can have overlapping fiscal years across companies they own/are
member of.

Also adds scripts/seed-demo-account.ts for end-to-end demo seeding
(two companies, full FY2025, active FY2026 with mixed state).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* fix(pr-402): address review feedback

Migrations
- Drop 20260506140000_document_journal_entry_immutability_delete_bypass.sql:
  redundant with 20260506140000_document_journal_entry_immutability_bypass.sql
  that landed on main while this branch was open. Both share the same
  gnubok.allow_delete pattern; main's version is what the DB actually has.
- Rename 20260506150000_bankid_enrichment_table.sql →
  20260506160000_bankid_enrichment_table.sql to clear the timestamp clash
  with 20260506150000_protect_document_journal_link.sql on main (Supabase
  branch preview was failing on schema_migrations PK collision).

Tests
- Drop the swap-under-flag test from delete-last-voucher.pg.test.ts:
  main's bypass returns NEW unconditionally when gnubok.allow_delete='true',
  so the swap is permitted. Drop the duplicate happy-path test (already
  covered by 'clears journal_entry_id on attached documents and deletes
  the voucher'). Keep the unauthorized-direct-UPDATE test.
- Add bankid-enrichment.pg.test.ts covering the SELECT RLS policy:
  user reads own row, cannot read another user's row, INSERT denied for
  authenticated.

gnubok_query_journal
- amount_min/amount_max is applied post-fetch (PostgREST can't OR
  abs(debit) and abs(credit) cleanly), but PostgREST's count is computed
  pre-filter. Reporting that as total_lines mislead agents into
  paginating a tail that was already filtered out. When the amount
  filter is applied, anchor total_lines and truncated to the filtered
  set and surface db_matched_pre_amount_filter +
  amount_filter_applied_post_fetch separately.
- Escape `_` in the free-text LIKE filter so a search for "2_441"
  doesn't match "2X441".

VAT close check
- Reverse-charge blocker no longer fires on ruta 30 (seller-side
  domestic omvänd skattskyldighet) — the seller books no VAT, the buyer
  does, so missing ruta 48 is expected. Now scoped to ruta 31/32 (EU
  acquisition) where the buyer must book both calculated output (2615)
  and matching ingående moms (2645).
- High-value receipt threshold no longer reads journal_entries.total_amount
  (column doesn't exist; check silently never fired). Sums debits across
  the entry's lines, which equals the gross for ordinary purchase entries
  — comparing a gross figure against the BFL/ML 4 000 SEK threshold per
  ML 17 kap 26–28 §.

seed-demo-account.ts
- Require an explicit email argument; refuse to run with the previously
  hardcoded fallback that would silently target a real user. Ensure
  email is non-undefined for downstream typing.
- Type the supabase fiscal_periods insert result locally so tsc no longer
  reports 'fp implicitly any' from the loose untyped client.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* fix(test): adjust fiscal-period-start-day pg test for per-company overlap

The pg-real failure on PR #402 was a latent bug surfaced by this branch's
fiscal_periods exclusion constraint flip from user_id to company_id
(migration 20260506140100). The test was inserting periods that overlapped
seedCompany's default 2026-01-01..2026-12-31 period; the previous
constraint slipped past it because the test's INSERT didn't set user_id
(NULL escapes the WITH = match), so two same-company overlapping periods
silently coexisted.

Now that the constraint correctly fires per company, pick years that
don't overlap with the seeded 2026 period. The trigger's behavior under
test (allow mid-month start when no earlier period exists, allow
back-dated SIE imports, reject mid-month start when an earlier period
exists) is unchanged.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* fix(vat-close-check): correct reverse-charge/import blocker rutor

Rutor 30/31/32 are the buyer's calculated utgående moms on reverse-
charge purchases (domestic byggtjänster/electronics → 2614 → ruta 30;
EU goods → 2624 → ruta 31; EU services → 2634 → ruta 32). The buyer
must also book matching ingående moms (2647 inhemskt / 2645 utlandet
→ ruta 48). The previous fix removed ruta 30 on the basis that it was
seller-side; that's incorrect — domestic-RC sellers book no VAT at
all (they report only beskattningsunderlag on ruta 41), so 2614 only
sees buyer-side entries. Restore ruta 30.

Also extend the check to import rutor 60/61/62 (non-EU import VAT
declared via momsdeklaration since 2015 — 2615/2625/2635). Same
mechanic: importer books output VAT on these rutor and deducts the
input side via ruta 48. SaaS-from-AWS / OpenAI / Vercel companies hit
this path; without including 60/61/62 the blocker would silently miss
their misbookings.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* fix(mcp): expose ruta 60/61/62 (import VAT) on the local VatReportResult

The vat-close-check fix referenced vatReport.rutor.ruta60/61/62 but the
MCP server's local VatReportResult type only carries ruta 05-49. Build
broke on tsc.

Extend the MCP server's slim VAT report to also project import VAT —
2615 → ruta 60 (25%), 2625 → ruta 61 (12%), 2635 → ruta 62 (6%) — and
fold those into ruta 49 (att betala/återfå). Mirrors the BAS-to-Ruta
mapping in lib/reports/vat-declaration.ts. Output schema and required
list updated accordingly.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-05-06 16:41:36 +02:00