Jakob WennbergandClaude Fable 5
27ef398623
With deferred tool loading on the client side (Claude Code ToolSearch, claude.ai connector search), an agent starting a bookkeeping session burns 4-6 round-trips discovering tools cluster by cluster. The briefing now returns recommended_tools: five per-workflow loadouts (categorize_month, close_period, invoice_run, vat_declaration, payroll_month), each with a stable key, one-line description, the gnubok_load_skill slug for the full playbook, and the exact registry tool names ordered by typical call sequence, so a batch-selection harness loads a whole workflow cluster in ONE ToolSearch select call. Drift protection: assertRecommendedLoadoutsValid() runs at module init in server.ts right after the tools array is built, failing module load (and every test importing the server) if a loadout names a tool absent from the registry or a skill slug absent from workflowSkills; the agent-briefing test suite pins the same checks. The list is static per company: the briefing does not query workflow state today, so gating inclusion would add reads to the bootstrap hot path. tools/list payload ceiling bumped 57K to 57.5K with a documented progression entry (schema prose trimmed to the floor first; headroom before the change was ~15 tokens). Fixes #1098 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