Jakob WennbergandClaude Fable 5
4a0b524fbb
* fix(categorization): connect card descriptors to counterparty history
suggest_categories returned no signal for recurring card merchants
(reported: Anthropic booked to 5420 fourteen times, zero suggestions).
Three compounding causes, all fixed:
- normalizeCounterpartyName() now reduces card-network descriptors to
their merchant segment ("ANTHROPIC* CLAUDE SUB SAN FRANCISCO" ->
"anthropic"; "PAYPAL *SPOTIFY" -> "spotify"), so monthly per-charge
tails stop splintering one merchant into unmatchable variants. SQL
mirror normalize_counterparty_key() updated in lockstep (migration
20260721140000), keeping the ledger-context template join exact.
- New token_subset match tier bridges templates learned from manual
bookings ("Claude Dec" -> "claude") to bank descriptors containing
the token, and card-core descriptors to legacy splintered templates.
Guarded by a distinctive-token filter so generic/geo words never
match on their own.
- Merchant history falls back to description when merchant_name is
null: card purchases never carry merchant_name, so the history path
was structurally blind to exactly the transactions that need it.
History keys now share the counterparty-template normalization and
the 200-row window is ordered by recency.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(categorization): guard single-token matches, anchor history on original_description
Review follow-ups (CodeRabbit on #1095):
- token_subset tier: a single shared distinctive token now also requires
occurrence_count >= 3 on the template, so a template named after a
common word or first name (one prior booking) cannot vacuum up
unrelated transfers ("SWISH ANDERS JOHANSSON"). Multi-token agreement
stays unrestricted; the Claude/Anthropic case (14 bookings) is
unaffected.
- merchant history keys on original_description ?? description: the raw
bank descriptor is immutable while description is a user-editable
working title, so renaming a transaction no longer severs its history
link for future recurring charges.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore(migrations): re-timestamp card-descriptor migration after prod moved past it
Prod applied 20260721144311 (#1101) through 20260721201747 (#1104) while
this PR was open; 20260721140000 would sort before them and risk being
skipped by out-of-order auto-apply at merge. Not yet applied to prod, so
renaming is safe; the preview branch re-applies idempotently
(CREATE OR REPLACE).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Accounted MCP server
JSON-RPC 2.0 server exposing the Accounted bookkeeping engine to MCP clients (Claude Desktop, Claude Code, etc.). Endpoint: /api/extensions/ext/mcp-server/mcp. OAuth and stdio bridge live alongside the API surface: see app/api/mcp-oauth/ and packages/gnubok-mcp/.
Tool authoring contract
Enforced by tests in __tests__/: these are not style preferences, they're guard rails.
additionalProperties: falseon everyinputSchema. Guarded bystrict-schemas.test.ts. Forces clear rejections on hallucinated fields instead of silent ignores.- Descriptions ≤ 280 chars. Guarded by
output-schema.test.ts. NoArgs:/Returns:/Examples:prose: those belong in JSON Schema. Use agent-native hints ("Use to…", "Call X first", "HIGH risk"). - Staged-operation envelope for write tools:
outputSchema: STAGED_OPERATION_SCHEMA(server.ts). Fields:staged, risk_level, actor, message, preview, period_status?, next?. Thestaged: trueboolean is the explicit completion signal; agents must not infer completion from prose. Do NOT introduce a parallel{ success, shouldContinue, output }envelope. period_statusthreading: any tool that ties to a fiscal-period-bound date (categorize, mark paid, create voucher, correct/reverse entry, approve supplier invoice) passesdateForPeriodChecktostagePendingOperation. Response then includesperiod_status: { period_id, status: open|locked|closed, lock_date }so widgets and agents disable writes without round-trips.- Scope mapping: every new tool needs an entry in
lib/auth/api-keys.tsTOOL_SCOPE_MAP. Missing entries default to deny. - Tests for new write tools: add staging-gate coverage to
__tests__/voucher-tools.test.ts(or a sibling) plus executor coverage tolib/pending-operations/__tests__/voucher-executors.test.tsif the tool stages a newoperation_type.
Determinism / cache stability
Tool definitions (name, description, inputSchema, outputSchema, annotations) are declared as static object literals at module load: no timestamps, no UUIDs, no Date/Math.random in the definition layer. This makes the tools/list JSON payload byte-stable across requests, which lets agent-side prompt caches stay warm. Do not introduce per-request non-determinism into the definitions block. Anything time-bound or random belongs inside execute().
For internal Anthropic API usage (today only extensions/general/invoice-inbox/lib/extract-invoice-fields.ts): annotate stable prefixes with cache_control: { type: 'ephemeral' } and log usage.cache_read_input_tokens for hit-ratio observability. The 1h TTL from the agent-native API plan (item 10) requires the direct Anthropic API; Accounted's Bedrock path defaults to a shorter TTL.
Payload-size watchdog
payload-size.bench.test.ts enforces a tools/list JSON payload ceiling. If the test fires, the right answer is rarely "raise the ceiling". Instead, trim descriptions or set specialized wide tools to catalogVisibility: 'search'. Those tools remain discoverable with full schemas through gnubok_search_tools and callable through tools/call without bloating the default catalog.
Where things live
server.ts: the tools array + JSON-RPC dispatchertool-result.ts:withNext(),toToolError()response helpersresources/: read-onlyAccounted://URIs (active company, period, recent activity, capabilities, attention items, voucher gaps, chart of accounts, VAT treatments)widgets/: inline HTML widgets (receipt-matcher, vat-review)prompts/: slash-command-style promptsskills/: domain-knowledge skill bodies served viagnubok_load_skill__tests__/: strictness guards + per-tool coverage