logicsrc/packages/logicsrc-mcp/src/openprd.ts
Anthony Ettinger da5f6f8381
Some checks are pending
CI / build (push) Waiting to run
test / test (push) Waiting to run
feat(openontology): Phase 2 + Phase 3 — storage, REST/SSE, MCP, RDF/SHACL, adapters, TUI, explorer (#101)
Everything the two shipped PRD phases deferred, minus what is called out below.

Storage (Phase 2)
  @logicsrc/openontology gains a SQLite/Turso adapter. It hydrates the read
  model at open, serves reads synchronously — a query evaluator that awaits per
  triple pattern is unusable — and buffers mutations as SQL that flush() writes
  in one transaction. Versioned idempotent migrations; indexes over subject,
  predicate, entity-valued object, status, both time axes, aliases, and external
  ids; FTS5 for label/alias search. The append-only status log is replayed on
  open, so retractions, supersessions, and merge redirects survive a reopen.

REST + SSE + OpenAPI (Phase 2)
  16 paths under /api/ontologies in logicsrc-web, described at
  /api/ontologies/openapi and referencing the published JSON Schemas rather
  than restating them. No token is read-only; a curator token can apply; an
  agent token can propose and cannot apply. Idempotency-Key on mutations,
  revision ETags, 409 on a stale base revision, and an SSE stream that emits
  the same event objects as the JSON endpoint.

MCP (Phase 2)
  OpenOntology and OpenPRD surfaces on the standards server: spec/manifest/
  schema/queries and PRD spec/index as resources, 11 ontology tools and 6 PRD
  tools, 7 prompts. Read-only by default; OPENONTOLOGY_MCP_WRITABLE=1 buys
  proposals, never applies — the denial is the shared policy layer, not a
  second rule that could drift.

Interoperability (Phase 3)
  RDF/Turtle export and import of the reified profile, plus the plain triple
  for asserted relationships so a consumer wanting only the accepted graph gets
  one. SHACL for 5 of 7 constraint kinds; `unique` and `query` are reported as
  unmapped in both the return value and the generated Turtle, because a shape
  that quietly means something narrower is worse than no shape.

Source adapters (Phase 3)
  CSV, JSON, YAML, NDJSON, Markdown, generic JSON HTTP, and GitHub. All produce
  PROPOSED change-set operations with source, evidence selector, run id, and
  confidence attached; fetch is injected so ingestion is offline and testable.
  Each declares its capabilities, so "nothing was deleted upstream" is never
  confused with "this adapter cannot see deletions" — none of the seven can.

TUI + explorer
  Keyboard-first panels (types, entities, claims, sources, queries, change
  sets, validation, audit) as plain strings that survive SSH and 60 columns;
  status is a glyph and a word, never colour alone; the key bar wraps rather
  than truncating. Wired as `logicsrc ontology tui`. A read-only web explorer
  at /openontology/explore with entity and claim views showing status, both
  clocks, confidence, sources, evidence, and append-only history — plus an
  /openprd page for the companion standard.

Bugs found and fixed while testing
  - the API built a new engine per request, so `explain` could never find a
    resultId from a prior request; engines are now cached per role
  - the TUI status bar called engine.validateOntologyPackage(), appending a
    package.validated event on every repaint; it now uses the pure validator

Verification: 76 new tests (527 total across the monorepo, all passing); full
build green; the libSQL adapter is exercised against real files, the API
through its route handlers, and MCP over an in-memory transport.

Not included: PWA review/approval write flows (they need an auth story this
deployment does not have), OWL/RDFS mappings, SPARQL/Cypher/Datalog query
adapters, and Phase 4 governed actions. The compatibility matrix marks those
"planned", not "supported".

Refs: prd/0001-add-logicsrc-openontology-spec.md

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-28 05:10:33 -07:00

278 lines
9.5 KiB
TypeScript

import { existsSync } from "node:fs";
import { resolve } from "node:path";
import type { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { z } from "zod";
import {
findPrd,
loadPrdCollection,
nextPrdNumber,
nextStatuses,
prdToTasks,
renderDocument,
renderIndex,
renderReport,
summarize,
validatePrdCollection,
validateTasks,
type PrdCollection
} from "@logicsrc/openprd";
/**
* OpenPRD surface for the MCP server.
*
* Everything here is read-only. Creating or moving a PRD writes to a repo, and
* that belongs to the CLI where a human sees the diff — not to a tool an agent
* can call unattended.
*/
const SPEC = `# OpenPRD
A lightweight standard for product requirements documents. A repo keeps a
numbered, committed collection under prd/, one Markdown file each.
- One file per PRD: prd/<id>-<slug>.md, four-digit ids, no gaps, 0000 reserved
for the template.
- Front-matter carries openprd, id, title, status, authors, and optional repo,
dates, discussion, implementation, tags, supersedes, superseded-by.
- The body has eight required sections in order: Problem, Goals, Non-Goals,
Users, Requirements, UX Notes, Success Metrics, Risks & Open Questions.
- Requirements are numbered R1, R2, … each tagged [P0], [P1], or [P2].
- Lifecycle: Draft → Review → Accepted → Final, or Rejected / Withdrawn /
Superseded. Status lives in front-matter and is the source of truth.
Full specification: https://logicsrc.com/docs/openprd`;
function textResult(text: string) {
return { content: [{ type: "text" as const, text }] };
}
function errorResult(text: string) {
return { content: [{ type: "text" as const, text }], isError: true };
}
function openCollection(): { collection: PrdCollection | null; dir: string | null; error: string | null } {
const dir = process.env.OPENPRD_DIR ?? "./prd";
const path = resolve(dir);
if (!existsSync(path)) {
return { collection: null, dir: path, error: `${path} does not exist; set OPENPRD_DIR` };
}
try {
return { collection: loadPrdCollection(path), dir: path, error: null };
} catch (error) {
return { collection: null, dir: path, error: (error as Error).message };
}
}
export function registerOpenPrd(server: McpServer): void {
server.registerResource(
"openprd-spec",
"logicsrc://openprd/spec",
{
title: "OpenPRD specification",
description: "Numbered PRDs: layout, front-matter, the eight sections, and the lifecycle.",
mimeType: "text/markdown"
},
async () => ({
contents: [{ uri: "logicsrc://openprd/spec", mimeType: "text/markdown", text: SPEC }]
})
);
server.registerResource(
"openprd-index",
"prd://index",
{
title: "PRD index",
description: "The current collection index, generated from the PRDs on disk.",
mimeType: "text/markdown"
},
async () => {
const { collection, error } = openCollection();
const text = collection ? renderIndex(collection) : `No PRD collection: ${error}`;
return { contents: [{ uri: "prd://index", mimeType: "text/markdown", text }] };
}
);
server.registerTool(
"prd_list",
{
title: "List PRDs",
description: "Lists the PRDs in the collection with id, title, status, tags, and requirement count.",
inputSchema: { status: z.string().optional().describe("Comma-separated statuses to include.") },
annotations: { readOnlyHint: true, openWorldHint: false }
},
async ({ status }) => {
const { collection, error } = openCollection();
if (!collection) return errorResult(error ?? "no collection");
const wanted = status?.split(",").map((value) => value.trim());
const rows = collection.documents
.map(summarize)
.filter((row) => !wanted || wanted.includes(row.status));
return textResult(JSON.stringify(rows, null, 2));
}
);
server.registerTool(
"prd_show",
{
title: "Show a PRD",
description: "Shows one PRD's front-matter, sections, and parsed requirements.",
inputSchema: {
ref: z.string().describe("Id, number, slug, or filename."),
format: z.enum(["text", "json", "markdown"]).optional()
},
annotations: { readOnlyHint: true, openWorldHint: false }
},
async ({ ref, format }) => {
const { collection, error } = openCollection();
if (!collection) return errorResult(error ?? "no collection");
const doc = findPrd(collection, ref);
if (!doc) return errorResult(`No PRD matching "${ref}"`);
return textResult(renderDocument(doc, format ?? "text"));
}
);
server.registerTool(
"prd_validate",
{
title: "Validate the PRD collection",
description:
"Checks conformance — filename, front-matter, id match, the eight sections in order — plus collection rules.",
inputSchema: { strict: z.boolean().optional() },
annotations: { readOnlyHint: true, openWorldHint: false }
},
async ({ strict }) => {
const { collection, error } = openCollection();
if (!collection) return errorResult(error ?? "no collection");
const report = validatePrdCollection(collection, {
strict: strict === true,
expectedIndex: renderIndex(collection)
});
return textResult(renderReport(report, "markdown"));
}
);
server.registerTool(
"prd_next_id",
{
title: "Next free PRD id",
description: "Returns the next four-digit id. Numbers are assigned at creation, never reserved.",
annotations: { readOnlyHint: true, openWorldHint: false }
},
async () => {
const { collection, error } = openCollection();
if (!collection) return errorResult(error ?? "no collection");
return textResult(nextPrdNumber(collection));
}
);
server.registerTool(
"prd_next_statuses",
{
title: "Allowed lifecycle moves",
description: "Given a PRD, lists the statuses it may legally move to next.",
inputSchema: { ref: z.string() },
annotations: { readOnlyHint: true, openWorldHint: false }
},
async ({ ref }) => {
const { collection, error } = openCollection();
if (!collection) return errorResult(error ?? "no collection");
const doc = findPrd(collection, ref);
if (!doc) return errorResult(`No PRD matching "${ref}"`);
const allowed = nextStatuses(doc.frontMatter.status);
return textResult(
JSON.stringify(
{
id: doc.frontMatter.id,
status: doc.frontMatter.status,
allowedNext: allowed,
terminal: allowed.length === 0
},
null,
2
)
);
}
);
server.registerTool(
"prd_tasks",
{
title: "Map requirements to LogicSRC tasks",
description: "Turns each R# into one logicsrc.task document, validated before it is returned.",
inputSchema: {
ref: z.string(),
priority: z.string().optional().describe("Comma-separated priorities, e.g. P0,P1"),
creator: z.string().optional()
},
annotations: { readOnlyHint: true, openWorldHint: false }
},
async ({ ref, priority, creator }) => {
const { collection, error } = openCollection();
if (!collection) return errorResult(error ?? "no collection");
const doc = findPrd(collection, ref);
if (!doc) return errorResult(`No PRD matching "${ref}"`);
const priorities = priority?.split(",").map((value) => value.trim()) as
| Array<"P0" | "P1" | "P2">
| undefined;
const { tasks, skipped } = prdToTasks(doc, { creator, priorities });
const problems = validateTasks(tasks);
if (problems.length > 0) {
return errorResult(`Generated tasks failed validation: ${JSON.stringify(problems, null, 2)}`);
}
return textResult(JSON.stringify({ tasks, skipped }, null, 2));
}
);
server.registerPrompt(
"write_prd",
{ title: "Draft an OpenPRD document", description: "Draft a conforming PRD for a product decision." },
async () => ({
messages: [
{
role: "user" as const,
content: {
type: "text" as const,
text: `Draft an OpenPRD document for the change the user describes.
Front-matter: openprd "0.2", a four-digit id matching the filename, an imperative
title starting with a verb, status Draft, and at least one author.
Then all eight sections, in this order, none omitted:
Problem, Goals, Non-Goals, Users, Requirements, UX Notes, Success Metrics,
Risks & Open Questions. A section may be a single line such as _None._
Requirements are numbered R1, R2, … contiguously, each tagged [P0], [P1], or [P2],
one capability per line. Goals are outcomes, not features. Non-Goals bound the work.
Risks & Open Questions must name the decisions still owed rather than pretending
they are settled.`
}
}
]
})
);
server.registerPrompt(
"review_prd",
{ title: "Review a PRD", description: "Review a PRD for shape, clarity, and honesty." },
async () => ({
messages: [
{
role: "user" as const,
content: {
type: "text" as const,
text: `Review this PRD.
Check the shape first: all eight sections present and in order, requirements numbered
contiguously with priority tags, front-matter complete.
Then the substance: are the Goals outcomes rather than features? Do the Non-Goals
actually bound the work? Is every P0 requirement testable? Do the Success Metrics
measure the Goals? Do the Risks name real decisions still owed, or is that section
decoration? Say what you would change and why.`
}
}
]
})
);
}