@cyanheads/brapi-mcp-server 0.7.11 → 0.7.13

package/AGENTS.md

@@ -0,0 +1,453 @@
+# Agent Protocol
+
+**Server:** brapi-mcp-server
+**Version:** 0.7.13
+**Framework:** [@cyanheads/mcp-ts-core](https://www.npmjs.com/package/@cyanheads/mcp-ts-core) `^0.13.6`
+**Engines:** Bun ≥1.4.0, Node ≥24.0.0
+**MCP SDK:** `@modelcontextprotocol/server` ^2.0.0 (protocol revisions 2026-07-28 and 2025-*)
+
+> **Read the framework docs first:** `node_modules/@cyanheads/mcp-ts-core/CLAUDE.md` contains the full API reference — builders, Context, error codes, exports, patterns. This file covers server-specific conventions only.
+
+---
+
+## What's Next?
+
+When the user asks what to do next, what's left, or needs direction, suggest relevant options based on the current project state:
+
+1. **Re-run the `setup` skill** — ensures AGENTS.md, skills, structure, and metadata are populated and up to date with the current codebase
+2. **Run the `design-mcp-server` skill** — if the tool/resource surface hasn't been mapped yet, work through domain design
+3. **Add tools/resources/prompts** — scaffold new definitions using the `add-tool`, `add-app-tool`, `add-resource`, `add-prompt` skills
+4. **Add services** — scaffold domain service integrations using the `add-service` skill
+5. **Add tests** — scaffold tests for existing definitions using the `add-test` skill
+6. **Field-test definitions** — exercise tools/resources/prompts with real inputs using the `field-test` skill, get a report of issues and pain points
+7. **Run `devcheck`** — lint, format, typecheck, and security audit
+8. **Run the `security-pass` skill** — audit handlers for MCP-specific security gaps: output injection, scope blast radius, input sinks, tenant isolation
+9. **Run the `polish-docs-meta` skill** — finalize README, CHANGELOG, metadata, and agent protocol for shipping
+10. **Run the `maintenance` skill** — investigate changelogs, adopt upstream changes, and sync skills after `bun update --latest`
+
+Tailor suggestions to what's actually missing or stale — don't recite the full list every time.
+
+---
+
+## Core Rules
+
+- **Logic throws, framework catches.** Tool/resource handlers are pure — throw on failure, no `try/catch`. The framework catches, classifies, and formats. Default to typed contracts: declare `errors: [...]` and throw via `ctx.fail(reason, …)` so failures carry stable `data.reason` codes for agent-client routing. Fall back to error factories (`notFound()`, `validationError()`, etc.) only for services or when no contract entry fits.
+- **Use `ctx.log`** for request-scoped logging. No `console` calls.
+- **Use `ctx.state`** for tenant-scoped storage. Never access persistence directly.
+- **Need input the caller didn't supply?** `return ctx.requestInput(...)` and read `ctx.inputs` when the handler is re-entered. Never `await` for user input mid-handler.
+- **Secrets in env vars only** — never hardcoded.
+- **Close the loop on issues.** When implementing work tracked by a GitHub issue, comment on the issue with what landed and close it. Do both — a comment without a close leaves stale issues open; a close without a comment leaves no record of what shipped. The comment is for future readers — state the concrete changes, not the conversation that produced them.
+
+---
+
+## Patterns
+
+### Tool — connection bootstrap
+
+`brapi_connect` is the session handshake. It registers the BrAPI server under a named alias, forces a capability refresh, and inlines the full orientation envelope so one call orients the agent. `baseUrl` and `auth` are both `optional()` — when omitted, `resolveConnectInput` fills them from `BRAPI_<ALIAS>_*` then `BRAPI_DEFAULT_*` env vars, so credentials never enter the LLM context. Same envelope is available on-demand via `brapi_server_info`.
+
+```ts
+// src/mcp-server/tools/definitions/brapi-connect.tool.ts (abbreviated)
+import { tool, z } from '@cyanheads/mcp-ts-core';
+import { JsonRpcErrorCode } from '@cyanheads/mcp-ts-core/errors';
+import { resolveConnectInput } from '@/config/alias-credentials.js';
+import { ConnectAuthSchema } from '../shared/connect-auth-schema.js';
+
+export const brapiConnect = tool('brapi_connect', {
+  description: 'Connect to a BrAPI v2 server… baseUrl + auth fall back to BRAPI_<ALIAS>_* / BRAPI_DEFAULT_* env vars when omitted.',
+  annotations: { openWorldHint: true, readOnlyHint: false, idempotentHint: true },
+  errors: [
+    { reason: 'auth_token_exchange_failed', code: JsonRpcErrorCode.Forbidden,
+      when: 'SGN or OAuth token exchange against /token failed',
+      recovery: 'Verify the credentials and that the server exposes /token before retrying.' },
+    { reason: 'auth_no_access_token', code: JsonRpcErrorCode.Forbidden,
+      when: 'Token endpoint responded but did not return an access_token',
+      recovery: 'Confirm the credentials are valid and the IdP issues access tokens for this grant.' },
+  ] as const,
+  input: z.object({
+    baseUrl: z.string().url().optional().describe('Falls back to BRAPI_<ALIAS>_BASE_URL → BRAPI_DEFAULT_BASE_URL.'),
+    auth: ConnectAuthSchema.optional().describe('Falls back to env-derived credentials.'),
+    alias: z.string().regex(/^[a-zA-Z0-9_-]+$/).default('default'),
+  }),
+  output: OrientationEnvelopeSchema,
+  async handler(input, ctx) {
+    const resolved = resolveConnectInput(input.alias, { baseUrl: input.baseUrl, auth: input.auth });
+    const connection = await getServerRegistry().register(ctx, {
+      alias: input.alias, baseUrl: resolved.baseUrl, auth: resolved.auth,
+    });
+    await getCapabilityRegistry().invalidate(connection.baseUrl, ctx);
+    return buildOrientationEnvelope(ctx, connection, { registry: getCapabilityRegistry(), client: getBrapiClient() });
+  },
+  format: (result) => [{ type: 'text', text: formatOrientationEnvelope(result) }],
+});
+```
+
+### Tool — find with dataframe spillover
+
+`find_*` tools share a pattern: pull one page capped at `loadLimit`, compute distributions across the returned rows, and if the upstream total exceeds `loadLimit` materialize the full union as a canvas dataframe and return a handle. Spilled rows live in DuckDB only — there is no parallel JSON store. Canvas is mandatory: startup fails closed when `core.canvas` is undefined.
+
+```ts
+// src/mcp-server/tools/definitions/brapi-find-germplasm.tool.ts (abbreviated)
+export const brapiFindGermplasm = tool('brapi_find_germplasm', {
+  description:
+    'Find germplasm by name, synonym, accession, PUI, crop, or free-text. Spills to a canvas dataframe when the upstream total exceeds loadLimit — query with brapi_dataframe_query (SQL).',
+  annotations: { readOnlyHint: true, openWorldHint: true },
+  input: z.object({
+    alias: AliasInput,
+    names: z.array(z.string()).optional(),
+    crops: z.array(z.string()).optional(),
+    text: z.string().optional(),
+    loadLimit: LoadLimitInput,
+    extraFilters: ExtraFiltersInput,
+  }),
+  output: OutputSchema,
+  async handler(input, ctx) {
+    const connection = await getServerRegistry().get(ctx, input.alias ?? DEFAULT_ALIAS);
+    await getCapabilityRegistry().ensure(connection.baseUrl, { service: 'germplasm', method: 'GET' }, ctx);
+    const bridge = getCanvasBridge();
+
+    const filters = mergeFilters(/* named + extraFilters */, warnings);
+    const firstPage = await loadInitialPage(client, connection, '/germplasm', filters, loadLimit, ctx);
+
+    const { fullRows, dataframe } = await maybeSpill({
+      firstPage, client, connection, bridge,
+      path: '/germplasm', filters, source: 'find_germplasm', loadLimit, ctx,
+    });
+    return { /* results + distributions + refinementHint + dataframe? */ };
+  },
+  format: (result) => [{ type: 'text', text: renderFindResult(result) }],
+});
+```
+
+### Server config
+
+```ts
+// src/config/server-config.ts — lazy-parsed, separate from framework config
+import { z } from '@cyanheads/mcp-ts-core';
+import { parseEnvConfig } from '@cyanheads/mcp-ts-core/config';
+
+const ServerConfigSchema = z.object({
+  defaultBaseUrl: z.string().url().optional(),
+  loadLimit: z.coerce.number().int().positive().default(1_000),
+  maxConcurrentRequests: z.coerce.number().int().positive().default(4),
+  retryMaxAttempts: z.coerce.number().int().min(0).default(3),
+  datasetTtlSeconds: z.coerce.number().int().positive().default(86_400),
+  referenceCacheTtlSeconds: z.coerce.number().int().positive().default(3_600),
+  sessionIsolation: z.enum(['true', 'false']).default('true').transform((v) => v === 'true'),
+  // …see src/config/server-config.ts for the full schema
+});
+
+let _config: z.infer<typeof ServerConfigSchema> | undefined;
+export function getServerConfig() {
+  _config ??= parseEnvConfig(ServerConfigSchema, {
+    defaultBaseUrl: 'BRAPI_DEFAULT_BASE_URL',
+    loadLimit: 'BRAPI_LOAD_LIMIT',
+    maxConcurrentRequests: 'BRAPI_MAX_CONCURRENT_REQUESTS',
+    retryMaxAttempts: 'BRAPI_RETRY_MAX_ATTEMPTS',
+    datasetTtlSeconds: 'BRAPI_DATASET_TTL_SECONDS',
+    referenceCacheTtlSeconds: 'BRAPI_REFERENCE_CACHE_TTL_SECONDS',
+  });
+  return _config;
+}
+```
+
+`parseEnvConfig` maps Zod schema paths → env var names so validation errors name the actual variable (`BRAPI_LOAD_LIMIT`) rather than the internal path (`loadLimit`). It throws a `ConfigurationError` the framework catches and prints as a clean startup banner.
+
+**Per-alias credentials** live in `src/config/alias-credentials.ts`. `readAliasCredentials(alias)` reads `BRAPI_<ALIAS>_*` (uppercased, hyphens → underscores), `deriveAuthFromCredentials(creds)` derives the auth mode from which fields are set (USERNAME+PASSWORD → `sgn`; BEARER_TOKEN → `bearer`; API_KEY → `api_key`; OAUTH_CLIENT_ID+SECRET → `oauth2`; mixing families raises `ValidationError`), and `resolveConnectInput(alias, agentInput)` layers agent input → alias env → default env → no-auth fallback.
+
+---
+
+## Context
+
+Handlers receive a unified `ctx` object. Currently used surface:
+
+| Property | Description |
+|:---------|:------------|
+| `ctx.log` | Request-scoped logger — `.debug()`, `.info()`, `.notice()`, `.warning()`, `.error()`. Auto-correlates requestId, traceId, tenantId. Dual-sink: Pino **and** `notifications/message` to the client, so treat it as client-visible. |
+| `ctx.state` | Tenant-scoped KV — used by `ServerRegistry` (connection aliases), `CanvasBridge` (default canvas pointer + per-table provenance), and `CapabilityRegistry` (cached profiles). Spilled `find_*` rows live on the canvas (DuckDB), not in `ctx.state`. |
+| `ctx.sessionId` | Mcp-Session-Id (HTTP stateful/auto); `undefined` for stdio, stateless HTTP unless `exposeStatelessSessionId` is opted in, and every request on protocol revision 2026-07-28, which is session-less by design. Composed into `ServerRegistry.connKey` and `CanvasBridge.defaultCanvasKey` when `BRAPI_SESSION_ISOLATION=true` (default), so concurrent HTTP sessions in the same tenant don't share connection state or canvas. Discovery / scoping key on top of tenant-keyed state — not an authorization principal. |
+| `ctx.signal` | `AbortSignal` — threaded into every BrAPI HTTP call so client-side cancellation aborts the upstream request. |
+| `ctx.requestId` | Unique request ID — auto-attached to every `ctx.log` entry. |
+| `ctx.tenantId` | Tenant ID from JWT or `'default'` for stdio / HTTP+`auth=none` — outer scope on all `ctx.state` reads/writes. |
+| `ctx.requestInput` / `ctx.inputs` | Multi-round-trip input. `brapi_submit_observations` gates apply-mode writes on a `confirm` elicitation: it reads `ctx.inputs.view('confirm')`, returns `ctx.requestInput({ inputRequests: … })` when the answer is missing, and treats a declined, cancelled, or unparseable answer as terminal (`user_declined`). `force: true` skips the round. |
+| `ctx.enrich` | Success-path agent context. `brapi_dataframe_query` and `brapi_build_phenotype_matrix` disclose capped results with `ctx.enrich.truncated({ shown, cap, guidance })`. |
+
+`ctx.fail(reason, …)` is the typed thrower keyed off declared `errors[]` contracts — used by 14 tools and 1 resource today. `ctx.recoveryFor(reason)` resolves the matching contract entry's recovery hint into `data.recovery.hint` so it surfaces on the wire. `ctx.content` is unused — no tool emits media blocks outside `brapi_get_image`, which returns image bytes through its own output schema.
+
+---
+
+## Errors
+
+Handlers throw — the framework catches, classifies, and formats.
+
+**Default for new tools: typed error contract.** Declare `errors: [{ reason, code, when, recovery, retryable? }]` on `tool()` to receive a typed `ctx.fail(reason, …)` keyed by the declared reason union. TypeScript catches `ctx.fail('typo')` at compile time, `data.reason` is auto-populated for observability, and the linter enforces conformance against the handler body. The `recovery` field is required descriptive metadata (≥ 5 words, lint-validated); to surface it on the wire, spread `...ctx.recoveryFor('reason')` into `data` or pass an explicit `{ recovery: { hint: '...' } }` when runtime context matters. Baseline codes (`InternalError`, `ServiceUnavailable`, `Timeout`, `ValidationError`, `SerializationError`, `RequestCancelled`) bubble freely and don't need declaring. Live across the BrAPI surface today: `brapi_build_phenotype_matrix`, `brapi_dataframe_describe`, `brapi_dataframe_export`, `brapi_dataframe_query`, `brapi_describe_filters`, `brapi_export_genotype_matrix`, `brapi_find_genotype_calls`, `brapi_germplasm_performance`, `brapi_get_germplasm`, `brapi_get_image`, `brapi_get_study`, `brapi_raw_get`, `brapi_raw_search`, `brapi_submit_observations`, plus the `brapi://variable/{observationVariableDbId}` resource.
+
+```ts
+errors: [
+  { reason: 'unknown_alias', code: JsonRpcErrorCode.NotFound,
+    when: 'No connection registered for this alias',
+    recovery: 'Call brapi_connect with this alias before retrying.' },
+],
+async handler(input, ctx) {
+  const conn = registry.peek(input.alias);
+  if (!conn) throw ctx.fail('unknown_alias', `No connection for ${input.alias}`,
+    { ...ctx.recoveryFor('unknown_alias') });
+  // ...
+}
+```
+
+**Declare contracts inline on each tool, even when similar across tools.** The contract is part of the tool's documented public surface — reading one tool definition file should give the full picture (input, output, errors, handler, format). Don't extract a shared `errors[]` constant or contract module to deduplicate near-identical entries; per-tool repetition is the intended cost of locality, and dynamic `recovery` hints often need tool-specific runtime context anyway.
+
+**Fallback (no contract entry fits, services, prototype tools):** throw via factories or plain `Error`.
+
+```ts
+// Plain Error — framework auto-classifies from message patterns
+throw new Error('Item not found');           // → NotFound
+throw new Error('Invalid query format');     // → ValidationError
+
+// Error factories — explicit code, concise
+import { notFound, validationError, internalError, serviceUnavailable } from '@cyanheads/mcp-ts-core/errors';
+throw notFound('Item not found', { itemId });
+throw serviceUnavailable('API unavailable', { url }, { cause: err });
+
+// McpError — full control over code and data
+import { McpError, JsonRpcErrorCode } from '@cyanheads/mcp-ts-core/errors';
+throw new McpError(JsonRpcErrorCode.DatabaseError, 'Connection failed', { pool: 'primary' });
+```
+
+Available factories include `notFound`, `validationError`, `forbidden`, `unauthorized`, `serviceUnavailable`, `rateLimited`, `timeout`, `conflict`, `internalError`, `serializationError`, `databaseError`, `configurationError`, `invalidParams`, `invalidRequest`. See framework CLAUDE.md for the full auto-classification table and the `api-errors` skill for contract patterns.
+
+---
+
+## Structure
+
+```text
+src/
+  index.ts                                # createApp() entry point — registers 25 tools, 6 resources, 2 prompts; inits 7 services
+  config/
+    server-config.ts                      # BRAPI_* env vars (Zod schema, lazy-parsed)
+    alias-credentials.ts                  # Per-alias env-var resolution (BRAPI_<ALIAS>_*) for brapi_connect
+  services/
+    brapi-client/                         # HTTP client — retry, concurrency cap, async-search poll, private-IP guard, binary fetch, POST/PUT
+    brapi-dialect/                        # Per-server filter / payload adapters (spec, cassavabase) — translates plural→singular, drops searchText, declares known-dead POST /search routes; envelope surfaces id + source + disabled-search nouns
+    brapi-filters/                        # Static v2.1 filter catalog
+    canvas-bridge/                        # Default-canvas resolver (per-session when BRAPI_SESSION_ISOLATION=true; per-tenant otherwise), df_<uuid> table generator, provenance store
+    capability-registry/                  # Per-connection /serverinfo cache + call guard
+    ontology-resolver/                    # Free-text → ontology-term matcher for variables
+    reference-data-cache/                 # Programs / trials / locations / crops lookup cache
+    server-registry/                      # Alias → live connection map with auth resolution; session-scoped under BRAPI_SESSION_ISOLATION=true
+  mcp-server/
+    tools/
+      definitions/
+        brapi-connect.tool.ts             # Session bootstrap — auth, capability load, orientation envelope
+        brapi-server-info.tool.ts         # Orientation envelope on demand
+        brapi-describe-filters.tool.ts    # Static BrAPI v2.1 filter catalog lookup
+        brapi-find-studies.tool.ts        # find_* — studies, distributions + spillover
+        brapi-get-study.tool.ts           # get_* — study + FK resolution + companion counts
+        brapi-find-germplasm.tool.ts      # find_* — germplasm
+        brapi-get-germplasm.tool.ts       # get_* — germplasm + attributes + parents + companion counts
+        brapi-walk-pedigree.tool.ts       # BFS DAG walk (ancestors / descendants / both) with cycle detection
+        brapi-find-variables.tool.ts      # find_* — observation variables, free-text ranking via OntologyResolver
+        brapi-find-observations.tool.ts   # find_* — observation records
+        brapi-find-images.tool.ts         # find_* — image metadata
+        brapi-get-image.tool.ts           # Fetch image bytes inline (imagecontent → imageURL fallback)
+        brapi-find-locations.tool.ts      # find_* — locations, optional client-side bbox filter
+        brapi-find-variants.tool.ts       # find_* — variants, 1-based inclusive/exclusive genomic region
+        brapi-find-genotype-calls.tool.ts # Async-search genotype calls with maxCalls cap + dataframe spillover
+        brapi-dataframe-describe.tool.ts  # List / describe canvas dataframes with columns, row counts, provenance
+        brapi-dataframe-query.tool.ts     # Run SQL across canvas dataframes (SELECT only); typed columns response
+        brapi-dataframe-drop.tool.ts      # Drop a dataframe by name (opt-in via BRAPI_CANVAS_DROP_ENABLED)
+        brapi-dataframe-export.tool.ts    # Write CSV/Parquet/JSON to BRAPI_EXPORT_DIR (opt-in, stdio-only)
+        brapi-build-phenotype-matrix.tool.ts  # Germplasm × trait matrix from studies; materialized as canvas dataframe
+        brapi-germplasm-performance.tool.ts   # Per-variable aggregates (n, mean, median, sd) for a single germplasm
+        brapi-export-genotype-matrix.tool.ts  # Genotype calls → germplasm × variant dataframe + VCF-lite / PLINK serialization
+        brapi-submit-observations.tool.ts # Two-phase observation write — preview / apply (POST + PUT) behind a confirmation round trip
+        brapi-raw-get.tool.ts             # Last-resort GET passthrough with routing nudge
+        brapi-raw-search.tool.ts          # Last-resort POST /search passthrough with async polling
+      shared/
+        connect-auth-schema.ts            # Tagged-union auth input
+        orientation-envelope.ts           # Shared envelope builder + formatter
+        find-helpers.ts                   # Alias / loadLimit / extraFilters fragments, mergeFilters, maybeSpill, DataframeHandleSchema
+        raw-routing-hints.ts              # Routing nudges emitted by raw_get / raw_search when a curated tool exists
+        canvas-columns.ts                 # SQL-safe column-name sanitizer + variableLegend builder (shared by matrix tools)
+        observations.ts                   # Study-anchored observation pull shared by phenotype-matrix and germplasm-performance
+        genotype-calls.ts                 # Async genotype-call collector shared by find-genotype-calls and export-genotype-matrix
+    resources/
+      definitions/
+        brapi-server-info.resource.ts     # brapi://server/info — orientation envelope (default connection)
+        brapi-calls.resource.ts           # brapi://calls — raw capability profile
+        brapi-study.resource.ts           # brapi://study/{studyDbId} — single study with FKs
+        brapi-germplasm.resource.ts       # brapi://germplasm/{germplasmDbId} — single germplasm with attributes + parents
+        brapi-filters.resource.ts         # brapi://filters/{endpoint} — filter catalog
+        brapi-variable.resource.ts        # brapi://variable/{observationVariableDbId} — observation variable (trait, scale, method)
+    prompts/
+      definitions/
+        brapi-eda-study.prompt.ts         # EDA playbook for one study (orient → variables → coverage → outliers → report)
+        brapi-meta-analysis.prompt.ts     # Cross-study meta-analysis (resolve trait → discover studies → harmonize → summarize)
+```
+
+---
+
+## Naming
+
+| What | Convention | Example |
+|:-----|:-----------|:--------|
+| Files | kebab-case with suffix | `search-docs.tool.ts` |
+| Tool/resource/prompt names | snake_case | `search_docs` |
+| Directories | kebab-case | `src/services/doc-search/` |
+| Descriptions | Single string or template literal, no `+` concatenation | `'Search items by query and filter.'` |
+
+---
+
+## Skills
+
+Skills are modular instructions in `framework-skills/` at the project root. Read them directly when a task matches — e.g., `framework-skills/add-tool/SKILL.md` when adding a tool. Keep development skills out of root `skills/`: plugin hosts load that directory for installing agents.
+
+**Agent skill directory:** Copy skills into the directory your agent discovers (Claude Code: `.claude/skills/`, Codex: `.codex/skills/`, shared: `.agents/skills/`, others: equivalent). This makes skills available as context without needing to reference `framework-skills/` paths manually. After framework updates, run the `maintenance` skill — it re-syncs the agent directory automatically (Phase B).
+
+Available skills:
+
+| Skill | Purpose |
+|:------|:--------|
+| `setup` | Post-init project orientation |
+| `design-mcp-server` | Design tool surface, resources, and services for a new server |
+| `add-tool` | Scaffold a new tool definition |
+| `add-app-tool` | Scaffold an MCP App tool + paired UI resource |
+| `add-resource` | Scaffold a new resource definition |
+| `add-prompt` | Scaffold a new prompt definition |
+| `add-service` | Scaffold a new service integration |
+| `add-test` | Scaffold test file for a tool, resource, or service |
+| `field-test` | Exercise tools/resources/prompts with real inputs, verify behavior, report issues |
+| `tool-defs-analysis` | Read-only audit of MCP definition language across the surface — voice, leaks, defaults, recovery hints, output descriptions |
+| `security-pass` | Audit server for MCP-flavored security gaps: output injection, scope blast radius, input sinks, tenant isolation |
+| `code-simplifier` | Post-session cleanup against `git diff` — modernize syntax, consolidate duplication, align with the codebase |
+| `techniques` | Catalog of response/data-shaping techniques — overflow handling, payload shaping, retrieval patterns |
+| `devcheck` | Lint, format, typecheck, audit |
+| `polish-docs-meta` | Finalize docs, README, metadata, and agent protocol for shipping |
+| `git-wrapup` | Land working-tree changes as a versioned commit stack; opens a release PR when the project declares release PR mode. |
+| `release-pr-review` | Review an open release PR, land fixups, and keep its body current. Release PR mode only. |
+| `release-and-publish` | Tag + push + npm + MCP Registry + GH Release + Docker. Picks up from `git-wrapup`. |
+| `maintenance` | Investigate changelogs, adopt upstream changes, sync skills to agent dirs |
+| `orchestrations` | Chain task skills into a gated multi-phase pipeline — build-out, QA-fix, update-ship — when you can spawn sub-agents |
+| `report-issue-framework` | File a bug or feature request against `@cyanheads/mcp-ts-core` via `gh` CLI |
+| `report-issue-local` | File a bug or feature request against this server's own repo via `gh` CLI |
+| `api-auth` | Auth modes, scopes, JWT/OAuth |
+| `api-linter` | Definition lint rule reference (`format-parity`, `schema-*`, `name-*`, `server-json-*`, …) |
+| `api-canvas` | DataCanvas: register tabular data, run SQL, export, plus the `spillover()` helper for big result sets — Tier 3 opt-in |
+| `api-config` | AppConfig, parseConfig, env vars |
+| `api-context` | Context interface, RequestContext, logger, state, multi-round-trip input |
+| `api-errors` | McpError, JsonRpcErrorCode, error patterns |
+| `api-mirror` | MirrorService: persistent SQLite-backed local mirror of a bulk upstream dataset with FTS5 — Tier 3 opt-in |
+| `api-services` | LLM, Speech, Graph services |
+| `api-telemetry` | OTel catalog: spans, metrics, completion logs, env config, cardinality rules |
+| `api-testing` | createMockContext, test patterns |
+| `api-utils` | Formatting, parsing, security, pagination, scheduling, telemetry helpers |
+| `api-workers` | Cloudflare Workers runtime |
+
+**Chaining skills into pipelines.** When the user wants a multi-phase effort — build this server out, QA-and-fix the surface, update-and-ship — *and you can spawn sub-agents*, `framework-skills/orchestrations/SKILL.md` sequences the task skills above into a gated pipeline with verification at each step. Read it to drive the run. Optional: skip it if you can't orchestrate sub-agents, and ignore it entirely if you were *spawned* as one — you've already been scoped to a single phase.
+
+When you complete a skill's checklist, check the boxes and add a completion timestamp at the end (e.g., `Completed: 2026-03-11`).
+
+---
+
+## Commands
+
+**Runtime:** Scripts and the production entry point use Bun directly — Bun executes TypeScript natively, no `tsx` shim. The `packageManager` field pins the version.
+
+| Command | Purpose |
+|:--------|:--------|
+| `bun run build` | Compile TypeScript |
+| `bun run rebuild` | Clean + build |
+| `bun run clean` | Remove build artifacts |
+| `bun run devcheck` | Lint + format + typecheck + security + changelog sync |
+| `bun run audit:fix` | Upgrade vulnerable packages within existing ranges with `bun audit fix`; try before `bun update <name>` and `bun dedupe`. |
+| `bun run audit:refresh` | Delete `bun.lock`, reinstall, re-audit. Last resort after in-place fixes; re-resolves ranged dependencies, including the framework. |
+| `bun run list-skills` | Print the skill index for this project (name, version, description) |
+| `bun run tree` | Generate `docs/tree.md` |
+| `bun run format` | Auto-fix formatting via Biome |
+| `bun run lint:mcp` | Validate MCP tool / resource / prompt definitions against the spec |
+| `bun run lint:packaging` | Verify env var alignment between `manifest.json` and `server.json` |
+| `bun run bundle` | Build, pack, and clean a `.mcpb` for one-click Claude Desktop install |
+| `bun run test` | Vitest suite |
+| `bun run start` | Production mode — defers transport selection to `MCP_TRANSPORT_TYPE` (stdio default) |
+| `bun run start:stdio` | Production mode (stdio) — requires prior `bun run build` |
+| `bun run start:http` | Production mode (HTTP) — requires prior `bun run build` |
+| `bun run changelog:build` | Regenerate `CHANGELOG.md` from `changelog/<minor>.x/` |
+| `bun run changelog:check` | Verify `CHANGELOG.md` is in sync (used by devcheck) |
+
+---
+
+## Bundling
+
+`bun run bundle` produces a `.mcpb` extension bundle for one-click install in Claude Desktop. The pack step is followed by `scripts/clean-mcpb.ts`, which prunes dev dependencies (`mcpb clean`) and strips two classes of `node_modules/**` content that root-anchored `.mcpbignore` patterns cannot reach: dependency-shipped agent docs (`framework-skills/`, `skills/`, `.claude/`, `.agents/`, `SKILL.md`) and platform-specific native bindings, which would otherwise lock the bundle to the platform it was packed on. The bundle therefore ships portable without the DuckDB native — `@duckdb/node-api` is loaded lazily, so canvas tools report an actionable install hint and every other tool works normally. MCPB is stdio-only — HTTP deployments are unaffected. To opt out, delete `manifest.json` and `.mcpbignore`; `lint:packaging` skips cleanly when `manifest.json` is absent.
+
+**Adding an env var requires both files:** `server.json` (registry discovery, `environmentVariables[]`) and `manifest.json` (bundle install UX, `mcp_config.env` + `user_config`). `lint:packaging` (run by `devcheck`) verifies the env var names match.
+
+---
+
+## Changelog
+
+Directory-based, grouped by minor series using the `.x` semver-wildcard convention. Source of truth is `changelog/<major.minor>.x/<version>.md` (e.g. `changelog/0.1.x/0.1.0.md`) — one file per released version, shipped in the npm package. At release time, author the per-version file with a concrete version and date, then run `npm run changelog:build` to regenerate the rollup. `changelog/template.md` is a **pristine format reference** — never edited, never renamed, never moved. Read it to remember the frontmatter + section layout when scaffolding a new per-version file. `CHANGELOG.md` is a **navigation index** (header + link + one-line summary per version), regenerated by `npm run changelog:build`. Devcheck hard-fails on drift. Never hand-edit `CHANGELOG.md`.
+
+Each per-version file opens with YAML frontmatter:
+
+```markdown
+---
+summary: One-line headline, ≤350 chars  # required — powers the rollup index
+breaking: false                          # optional — true flags breaking changes
+security: false                          # optional — true ONLY for a source-code security fix, never a dependency CVE bump
+---
+
+# 0.1.0 — YYYY-MM-DD
+...
+```
+
+`breaking: true` renders a `· ⚠️ Breaking` badge — use it when consumers must update code on upgrade (signature changes, removed APIs, config renames). `security: true` renders a `· 🛡️ Security` badge and pairs with a `## Security` body section — set it only for a security fix in this server's *own source code*, never for a routine dependency or transitive CVE bump (record those under `## Dependencies`). When both are set, badges render `· ⚠️ Breaking · 🛡️ Security`.
+
+`agent-notes` is an optional free-form field for maintenance agents processing the release downstream. Content here won't appear in the rendered CHANGELOG — it's consumed by agents running the `maintenance` skill. Use it for adoption instructions that don't fit the human-facing sections: new files to create, fields to populate, one-time migration steps. Omit entirely when there's nothing to say.
+
+**Section order** (Keep a Changelog): Added, Changed, Deprecated, Removed, Fixed, Security, then Dependencies. Include only sections with entries — don't ship empty headers.
+
+**Tag annotations** render as GitHub Release bodies via `--notes-from-tag`. Subject omits the version number (GitHub prepends it). Follow `framework-skills/release-and-publish/SKILL.md` for the tag format.
+
+---
+
+## Publishing
+
+**Every release goes through a release PR, straight-through** — `git-wrapup`'s "Release PR mode", mode `straight-through`. One run: `git-wrapup` lands the commit stack on `release/<version>`, pushes it, and opens the PR (title = the release commit subject, body = the changelog entry plus a gates section); `release-and-publish` then fast-forwards `main` locally with `git merge --ff-only`, creates the tag on `main`'s tip, pushes `main` and the tag, deletes the branch, and publishes. A caller's brief may run a given release as `gated` instead — a `release-pr-review` pass on the open PR before `release-and-publish`. **Never merge through the GitHub UI or `gh pr merge`**: squash and rebase-merge are disabled in the repo settings because both rewrite the stack (rebase-merge also strips the SSH signatures), and a merge commit breaks the linear history.
+
+---
+
+## Imports
+
+```ts
+// Framework — z is re-exported, no separate zod import needed
+import { tool, z } from '@cyanheads/mcp-ts-core';
+import { McpError, JsonRpcErrorCode } from '@cyanheads/mcp-ts-core/errors';
+
+// Server's own code — via path alias
+import { getMyService } from '@/services/my-domain/my-service.js';
+```
+
+---
+
+## Checklist
+
+- [ ] Zod schemas: all fields have `.describe()`, only JSON-Schema-serializable types (no `z.custom()`, `z.date()`, `z.transform()`, `z.bigint()`, `z.symbol()`, `z.void()`, `z.map()`, `z.set()`, `z.function()`, `z.nan()`)
+- [ ] Optional nested objects: handler guards for empty inner values from form-based clients (`if (input.obj?.field && ...)`, not just `if (input.obj)`). When schema-level regex/length matters, use `z.union([z.literal(''), z.string().regex(...).describe(...)])` — literal variants are exempt from `describe-on-fields`.
+- [ ] JSDoc `@fileoverview` + `@module` on every file
+- [ ] `ctx.log` for logging, `ctx.state` for storage — no `console`, no direct persistence access
+- [ ] Handlers throw on failure — error factories or plain `Error`, no try/catch
+- [ ] `format()` renders all data the LLM needs — different clients forward different surfaces (Claude Code → `structuredContent`, Claude Desktop → `content[]`); both must carry the same data
+- [ ] BrAPI tool: resolves connection via `ServerRegistry.get(ctx, alias ?? DEFAULT_ALIAS)` before touching the client
+- [ ] BrAPI tool: gates the call with `CapabilityRegistry.ensure(...)` — never fires against an endpoint the server didn't advertise
+- [ ] BrAPI tool: raw / domain / output schemas reviewed against real upstream sparsity (most `/germplasm` and `/studies` fields are optional in the wild)
+- [ ] BrAPI tool: normalization and `format()` preserve uncertainty — never fabricate missing IDs, names, or counts
+- [ ] BrAPI tool with dataframe spillover: rows beyond `loadLimit` materialize as a `df_<uuid>` canvas table via `CanvasBridge.registerDataframe`, handle surfaces in `result.dataframe`, `hasMore` set correctly
+- [ ] Tests include at least one sparse upstream payload (fields omitted) alongside the happy path
+- [ ] Registered in the `tools` array of `createApp()` in `src/index.ts`
+- [ ] Tests use `createMockContext()` from `@cyanheads/mcp-ts-core/testing`
+- [ ] `.codex-plugin/plugin.json` populated — `name`, `version`, `description`, `repository`, `license` from `package.json`; `interface.displayName` = the unscoped repo name; `interface.shortDescription` from `package.json` description
+- [ ] `.codex-plugin/mcp.json` updated — server name key is the unscoped repo name; user-supplied variables are listed in `env_vars`, never set to empty strings in `env`
+- [ ] `.claude-plugin/plugin.json` populated — metadata from `package.json`; inline `mcpServers` entry keyed by the unscoped repo name; user-supplied variables declared under `userConfig` and referenced as `${user_config.<option>}`
+- [ ] `bun run devcheck` passes

package/CLAUDE.md

@@ -1,9 +1,9 @@
 # Agent Protocol
 
 **Server:** brapi-mcp-server
-**Version:** 0.7.11
-**Framework:** [@cyanheads/mcp-ts-core](https://www.npmjs.com/package/@cyanheads/mcp-ts-core) `^0.12.3`
-**Engines:** Bun ≥1.3.0, Node ≥24.0.0
+**Version:** 0.7.13
+**Framework:** [@cyanheads/mcp-ts-core](https://www.npmjs.com/package/@cyanheads/mcp-ts-core) `^0.13.6`
+**Engines:** Bun ≥1.4.0, Node ≥24.0.0
 **MCP SDK:** `@modelcontextprotocol/server` ^2.0.0 (protocol revisions 2026-07-28 and 2025-*)
 
 > **Read the framework docs first:** `node_modules/@cyanheads/mcp-ts-core/CLAUDE.md` contains the full API reference — builders, Context, error codes, exports, patterns. This file covers server-specific conventions only.
@@ -180,7 +180,7 @@ Handlers receive a unified `ctx` object. Currently used surface:
 
 Handlers throw — the framework catches, classifies, and formats.
 
-**Default for new tools: typed error contract.** Declare `errors: [{ reason, code, when, recovery, retryable? }]` on `tool()` to receive a typed `ctx.fail(reason, …)` keyed by the declared reason union. TypeScript catches `ctx.fail('typo')` at compile time, `data.reason` is auto-populated for observability, and the linter enforces conformance against the handler body. The `recovery` field is required descriptive metadata (≥ 5 words, lint-validated); to surface it on the wire, spread `...ctx.recoveryFor('reason')` into `data` or pass an explicit `{ recovery: { hint: '...' } }` when runtime context matters. Baseline codes (`InternalError`, `ServiceUnavailable`, `Timeout`, `ValidationError`, `SerializationError`) bubble freely and don't need declaring. Live across the BrAPI surface today: `brapi_build_phenotype_matrix`, `brapi_dataframe_describe`, `brapi_dataframe_export`, `brapi_dataframe_query`, `brapi_describe_filters`, `brapi_export_genotype_matrix`, `brapi_find_genotype_calls`, `brapi_germplasm_performance`, `brapi_get_germplasm`, `brapi_get_image`, `brapi_get_study`, `brapi_raw_get`, `brapi_raw_search`, `brapi_submit_observations`, plus the `brapi://variable/{observationVariableDbId}` resource.
+**Default for new tools: typed error contract.** Declare `errors: [{ reason, code, when, recovery, retryable? }]` on `tool()` to receive a typed `ctx.fail(reason, …)` keyed by the declared reason union. TypeScript catches `ctx.fail('typo')` at compile time, `data.reason` is auto-populated for observability, and the linter enforces conformance against the handler body. The `recovery` field is required descriptive metadata (≥ 5 words, lint-validated); to surface it on the wire, spread `...ctx.recoveryFor('reason')` into `data` or pass an explicit `{ recovery: { hint: '...' } }` when runtime context matters. Baseline codes (`InternalError`, `ServiceUnavailable`, `Timeout`, `ValidationError`, `SerializationError`, `RequestCancelled`) bubble freely and don't need declaring. Live across the BrAPI surface today: `brapi_build_phenotype_matrix`, `brapi_dataframe_describe`, `brapi_dataframe_export`, `brapi_dataframe_query`, `brapi_describe_filters`, `brapi_export_genotype_matrix`, `brapi_find_genotype_calls`, `brapi_germplasm_performance`, `brapi_get_germplasm`, `brapi_get_image`, `brapi_get_study`, `brapi_raw_get`, `brapi_raw_search`, `brapi_submit_observations`, plus the `brapi://variable/{observationVariableDbId}` resource.
 
 ```ts
 errors: [
@@ -301,9 +301,9 @@ src/
 
 ## Skills
 
-Skills are modular instructions in `skills/` at the project root. Read them directly when a task matches — e.g., `skills/add-tool/SKILL.md` when adding a tool.
+Skills are modular instructions in `framework-skills/` at the project root. Read them directly when a task matches — e.g., `framework-skills/add-tool/SKILL.md` when adding a tool. Keep development skills out of root `skills/`: plugin hosts load that directory for installing agents.
 
-**Agent skill directory:** Copy skills into the directory your agent discovers (Claude Code: `.claude/skills/`, Codex: `.codex/skills/`, shared: `.agents/skills/`, others: equivalent). This makes skills available as context without needing to reference `skills/` paths manually. After framework updates, run the `maintenance` skill — it re-syncs the agent directory automatically (Phase B).
+**Agent skill directory:** Copy skills into the directory your agent discovers (Claude Code: `.claude/skills/`, Codex: `.codex/skills/`, shared: `.agents/skills/`, others: equivalent). This makes skills available as context without needing to reference `framework-skills/` paths manually. After framework updates, run the `maintenance` skill — it re-syncs the agent directory automatically (Phase B).
 
 Available skills:
 
@@ -324,8 +324,9 @@ Available skills:
 | `techniques` | Catalog of response/data-shaping techniques — overflow handling, payload shaping, retrieval patterns |
 | `devcheck` | Lint, format, typecheck, audit |
 | `polish-docs-meta` | Finalize docs, README, metadata, and agent protocol for shipping |
-| `git-wrapup` | Land working-tree changes as a versioned commit + annotated tag — version bump, changelog, verify, tag. Local only. |
-| `release-and-publish` | Push + npm + MCP Registry + GH Release + Docker. Picks up from `git-wrapup` |
+| `git-wrapup` | Land working-tree changes as a versioned commit stack; opens a release PR when the project declares release PR mode. |
+| `release-pr-review` | Review an open release PR, land fixups, and keep its body current. Release PR mode only. |
+| `release-and-publish` | Tag + push + npm + MCP Registry + GH Release + Docker. Picks up from `git-wrapup`. |
 | `maintenance` | Investigate changelogs, adopt upstream changes, sync skills to agent dirs |
 | `orchestrations` | Chain task skills into a gated multi-phase pipeline — build-out, QA-fix, update-ship — when you can spawn sub-agents |
 | `report-issue-framework` | File a bug or feature request against `@cyanheads/mcp-ts-core` via `gh` CLI |
@@ -343,7 +344,7 @@ Available skills:
 | `api-utils` | Formatting, parsing, security, pagination, scheduling, telemetry helpers |
 | `api-workers` | Cloudflare Workers runtime |
 
-**Chaining skills into pipelines.** When the user wants a multi-phase effort — build this server out, QA-and-fix the surface, update-and-ship — *and you can spawn sub-agents*, `skills/orchestrations/SKILL.md` sequences the task skills above into a gated pipeline with verification at each step. Read it to drive the run. Optional: skip it if you can't orchestrate sub-agents, and ignore it entirely if you were *spawned* as one — you've already been scoped to a single phase.
+**Chaining skills into pipelines.** When the user wants a multi-phase effort — build this server out, QA-and-fix the surface, update-and-ship — *and you can spawn sub-agents*, `framework-skills/orchestrations/SKILL.md` sequences the task skills above into a gated pipeline with verification at each step. Read it to drive the run. Optional: skip it if you can't orchestrate sub-agents, and ignore it entirely if you were *spawned* as one — you've already been scoped to a single phase.
 
 When you complete a skill's checklist, check the boxes and add a completion timestamp at the end (e.g., `Completed: 2026-03-11`).
 
@@ -359,7 +360,8 @@ When you complete a skill's checklist, check the boxes and add a completion time
 | `bun run rebuild` | Clean + build |
 | `bun run clean` | Remove build artifacts |
 | `bun run devcheck` | Lint + format + typecheck + security + changelog sync |
-| `bun run audit:refresh` | Delete `bun.lock`, reinstall, re-audit. Use when `devcheck` flags a transitive advisory — Bun's `update` is sticky on transitive resolutions, so the advisory may be a stale-lockfile false positive. If it survives the refresh, it's real. |
+| `bun run audit:fix` | Upgrade vulnerable packages within existing ranges with `bun audit fix`; try before `bun update <name>` and `bun dedupe`. |
+| `bun run audit:refresh` | Delete `bun.lock`, reinstall, re-audit. Last resort after in-place fixes; re-resolves ranged dependencies, including the framework. |
 | `bun run list-skills` | Print the skill index for this project (name, version, description) |
 | `bun run tree` | Generate `docs/tree.md` |
 | `bun run format` | Auto-fix formatting via Biome |
@@ -377,7 +379,7 @@ When you complete a skill's checklist, check the boxes and add a completion time
 
 ## Bundling
 
-`bun run bundle` produces a `.mcpb` extension bundle for one-click install in Claude Desktop. The pack step is followed by `scripts/clean-mcpb.ts`, which prunes dev dependencies (`mcpb clean`) and strips two classes of `node_modules/**` content that root-anchored `.mcpbignore` patterns cannot reach: dependency-shipped agent docs (`skills/`, `.claude/`, `.agents/`, `SKILL.md`) and platform-specific native bindings, which would otherwise lock the bundle to the platform it was packed on. The bundle therefore ships portable without the DuckDB native — `@duckdb/node-api` is loaded lazily, so canvas tools report an actionable install hint and every other tool works normally. MCPB is stdio-only — HTTP deployments are unaffected. To opt out, delete `manifest.json` and `.mcpbignore`; `lint:packaging` skips cleanly when `manifest.json` is absent.
+`bun run bundle` produces a `.mcpb` extension bundle for one-click install in Claude Desktop. The pack step is followed by `scripts/clean-mcpb.ts`, which prunes dev dependencies (`mcpb clean`) and strips two classes of `node_modules/**` content that root-anchored `.mcpbignore` patterns cannot reach: dependency-shipped agent docs (`framework-skills/`, `skills/`, `.claude/`, `.agents/`, `SKILL.md`) and platform-specific native bindings, which would otherwise lock the bundle to the platform it was packed on. The bundle therefore ships portable without the DuckDB native — `@duckdb/node-api` is loaded lazily, so canvas tools report an actionable install hint and every other tool works normally. MCPB is stdio-only — HTTP deployments are unaffected. To opt out, delete `manifest.json` and `.mcpbignore`; `lint:packaging` skips cleanly when `manifest.json` is absent.
 
 **Adding an env var requires both files:** `server.json` (registry discovery, `environmentVariables[]`) and `manifest.json` (bundle install UX, `mcp_config.env` + `user_config`). `lint:packaging` (run by `devcheck`) verifies the env var names match.
 
@@ -404,9 +406,15 @@ security: false                          # optional — true ONLY for a source-c
 
 `agent-notes` is an optional free-form field for maintenance agents processing the release downstream. Content here won't appear in the rendered CHANGELOG — it's consumed by agents running the `maintenance` skill. Use it for adoption instructions that don't fit the human-facing sections: new files to create, fields to populate, one-time migration steps. Omit entirely when there's nothing to say.
 
-**Section order** (Keep a Changelog): Added, Changed, Deprecated, Removed, Fixed, Security. Include only sections with entries — don't ship empty headers.
+**Section order** (Keep a Changelog): Added, Changed, Deprecated, Removed, Fixed, Security, then Dependencies. Include only sections with entries — don't ship empty headers.
 
-**Tag annotations** render as GitHub Release bodies via `--notes-from-tag`. They must be structured markdown — never a flat comma-separated string. Subject omits the version number (GitHub prepends it). See `changelog/template.md` for the full format reference.
+**Tag annotations** render as GitHub Release bodies via `--notes-from-tag`. Subject omits the version number (GitHub prepends it). Follow `framework-skills/release-and-publish/SKILL.md` for the tag format.
+
+---
+
+## Publishing
+
+**Every release goes through a release PR, straight-through** — `git-wrapup`'s "Release PR mode", mode `straight-through`. One run: `git-wrapup` lands the commit stack on `release/<version>`, pushes it, and opens the PR (title = the release commit subject, body = the changelog entry plus a gates section); `release-and-publish` then fast-forwards `main` locally with `git merge --ff-only`, creates the tag on `main`'s tip, pushes `main` and the tag, deletes the branch, and publishes. A caller's brief may run a given release as `gated` instead — a `release-pr-review` pass on the open PR before `release-and-publish`. **Never merge through the GitHub UI or `gh pr merge`**: squash and rebase-merge are disabled in the repo settings because both rewrite the stack (rebase-merge also strips the SSH signatures), and a merge commit breaks the linear history.
 
 ---
 
@@ -439,7 +447,7 @@ import { getMyService } from '@/services/my-domain/my-service.js';
 - [ ] Tests include at least one sparse upstream payload (fields omitted) alongside the happy path
 - [ ] Registered in the `tools` array of `createApp()` in `src/index.ts`
 - [ ] Tests use `createMockContext()` from `@cyanheads/mcp-ts-core/testing`
-- [ ] `.codex-plugin/plugin.json` populated — `name`, `version`, `description`, `repository`, `license` from `package.json`; `interface.displayName` = package name; `interface.shortDescription` from `package.json` description
-- [ ] `.codex-plugin/mcp.json` updated — server name key matches `package.json` name; env vars added for any required API keys
-- [ ] `.claude-plugin/plugin.json` populated — `name`, `version`, `description`, `repository`, `license` from `package.json`; inline `mcpServers` entry with server name key, env vars for any required API keys
+- [ ] `.codex-plugin/plugin.json` populated — `name`, `version`, `description`, `repository`, `license` from `package.json`; `interface.displayName` = the unscoped repo name; `interface.shortDescription` from `package.json` description
+- [ ] `.codex-plugin/mcp.json` updated — server name key is the unscoped repo name; user-supplied variables are listed in `env_vars`, never set to empty strings in `env`
+- [ ] `.claude-plugin/plugin.json` populated — metadata from `package.json`; inline `mcpServers` entry keyed by the unscoped repo name; user-supplied variables declared under `userConfig` and referenced as `${user_config.<option>}`
 - [ ] `bun run devcheck` passes