Files
rea/tests/boundary/mcp/analysisViewMcp.test.ts
3fca2e0ca6 feat(environment): extract the initial state a capture's reads observed (#5319)
* feat(environment): extract the initial state a capture's reads observed

Add the environment-state analysis view. From the parents the environment
manifest accepts, it extracts the data the capture's reads observed before
the first related write: entity types with an identity field (proved by a
path parameter or inferred from an identity name), entities with their
recorded values and provenance, documents for written non-entity paths, and
a render plan per read tool that is kept only when it reproduces the tool's
latest initial recordings from the state; every other read tool is a
recorded view with the reason. The Evidence is always inferred.

Response-body decoding moves to a domain owner shared with replay, exact
JSON parsing moves to the owner of the exact-number policy, the manifest and
state views share parent admission and record grouping, and
projectAnalysisView takes its accumulators as one object.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(environment): bound state at redacted writes; infer identity only from items

A scenario write whose URL was redacted has no request line, so it was
dropped and reads after it counted as initial state. Such a successful
write now joins the timeline with an empty path, whose root relates to
every type, so it ends every type's state; a limitation counts them.

A list whose recordings were all empty passed the "every listed item
carries it" check vacuously and published an identity field nothing
showed; it now needs at least one item, else the view stays recorded
as no-identity.

A value that differed between two request lines of one item tool, such
as a query variant, was called volatile, and the gate then substituted
each recording's own value; only repeats of one request line now show
volatility, so differing lines go through fields-differ.

List membership is compared only while a list can still be the whole
collection, and merged entities are indexed by type and collection once,
so paged captures no longer cost pages times entities (8,000 pages of
40,000 entities: 5.7 s to 0.23 s).

Docs and the tool contract say scenario state keeps the capture's
redaction markers and how redacted writes bound it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(environment): keep first sightings, one list pointer and linear field unions

Entities took the order of the first plan observed, and item plans run
before list plans, so a complete list read before its detail reads came
out reordered and lost "all" membership; each entity now keeps its
earliest recorded sighting.

A list plan reused the first recording's pointer for later recordings,
so a later response that was itself an entity wrapping a list was read
as that list; every recording must now locate the same list.

The field union scanned its ordered list for every key, quadratic in an
object's width; a set now answers membership.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(environment): keep post-write values and partial lists out of the initial state

A value that changed between repeats of one request line was named
volatile whenever no write under the type's own root lay between, so a
write under another root (for example POST /api/dashboards/db before a
second GET /api/search) left its effect in the initial state. Any
recorded write between the two observations now leaves every tool of the
type recorded with reason fields-differ instead.

A list was published with membership "all" even when a total or count
envelope member exceeded the number it listed. Such a counter now makes
the membership "recorded".

Also drop a stale doc comment on ObservedEntity.order.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* docs(environment): scope item-shaped and deleted written paths out of documents

State documents are held only for list-shaped paths a POST, PUT or PATCH
tool writes. Reads of item-shaped written paths and of paths that are only
deleted stay recorded views; the guide now says so and the spec defers them
to a later increment of #4933.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(environment): count writes in flight when a value changes between reads

The volatility check compared write start positions with the two reads,
so a write already in flight when the earlier read started was missed and
its effect was named volatile. A write now counts when its interval
overlaps the span from the earlier read's start to the later read's end.

The check also ran once per changed field and scanned every write. It now
runs once per pair of observations, answered by a binary search over the
writes in timeline order and the latest end among each prefix.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(environment): catch cross-tool changes across writes; read only item totals

A value that changed across a write was caught only between repeats of
one tool's request line, so a list read before a write and an item read
after it put the post-write value into the initial state. Changes are
now classified per entity in timeline order after observation: across
any recorded write, a change between repeats of a line or between two
lines of the type rewrites the type (every tool recorded as
fields-differ); with no write between, a repeat's change is volatile.
The classification lives in environmentStateMerge.ts.

Any numeric member whose name contained "total" or "count" counted as
an item total, so total_pages, account_id or page_count made complete
lists partial. Item totals are now members whose name tokens include
total or count and none of page, pages or per; a string of digits counts
as a number; and a set has_more, next or next-page token member also
marks the list partial.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(environment): keep post-write observations out of initial state

* fix(environment): retain field provenance and bound state inference work

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <gensokyo-dev@users.noreply.github.com>
2026-10-19 20:33:46 -07:00

1358 lines
45 KiB
TypeScript

import { advertisedMcpToolSchemas } from "../../fixtures/advertisedMcpTool.js";
import { API_SURFACE_EXAMPLE } from "../../../src/contracts/analysisViewApiSurfaceExample.js";
import { ENVIRONMENT_MANIFEST_EXAMPLE } from "../../../src/contracts/analysisViewEnvironmentManifestExample.js";
import { createHash } from "node:crypto";
import { readFile, writeFile } from "node:fs/promises";
import { WebNetworkCaptureService } from "../../../src/application/WebNetworkCaptureService.js";
import { decodeHarCapture } from "../../../src/browser/history/HarCapture.js";
import { join } from "node:path";
import {
apiSurfaceHar,
environmentManifestHar,
} from "../../fixtures/historicalHar.js";
import { createTestTempDirectory } from "../../fixtures/temporaryDirectory.js";
import { parseMcpToolError } from "../../fixtures/mcpToolError.js";
import { Client, InMemoryTransport } from "@modelcontextprotocol/client";
import { STDIO_DEFAULT_MAX_BUFFER_SIZE } from "@modelcontextprotocol/server";
import { Ajv2020 } from "ajv/dist/2020.js";
import { expect, it, onTestFinished } from "vitest";
import { z } from "zod";
import { BinaryLayoutService } from "../../../src/application/binaryDiagnostics/BinaryLayoutService.js";
import { toolContract } from "../../../src/contracts/toolContracts.js";
import { createEvidence, parseEvidence } from "../../../src/domain/evidence.js";
import { createAnalysisProfile } from "../../../src/domain/analysisProfile.js";
import { jsonObjectSchema } from "../../../src/domain/jsonValue.js";
import { functionDossierSchema } from "../../../src/domain/hopperValues.js";
import { ghidraFunctionDossier } from "../../../src/domain/ghidraValues.fixture.js";
import { ok } from "../../../src/domain/result.js";
import {
createJavaScriptApplicationGraph,
createJavaScriptApplicationNode,
} from "../../../src/domain/javascript/javascriptApplicationGraph.js";
import { createServer } from "../../../src/server/createServer.js";
import { connectGhidraMcp } from "./ghidraMcpHarness.js";
import {
analysisViewJavaScriptEvidence,
analysisViewJavaScriptAnalysisWithSource,
analysisViewBindJavaScriptGraphs,
analysisViewLayoutEvidence,
analysisViewLayoutFixture,
analysisViewCaptureFixture,
analysisViewCaptureEvidence,
analysisViewObjectDenseCapture,
analysisViewNativeEvidence,
environmentManifestCaptureFixture,
environmentStateCaptureFixture,
} from "../../fixtures/analysisView.js";
import { createTestBinarySession } from "../../fixtures/binarySession.js";
import { BINARY_LAYOUT_TEST_PROVIDER } from "../../fixtures/binaryDiagnostics/layout.js";
import { wasmArtifactFixture } from "../../fixtures/wasm/artifact.js";
import { evmInterfaceFixture } from "../../fixtures/evm/interface.js";
import type { RelationshipCoverage } from "../../../src/domain/relationshipCoverage.js";
const connect = async (
binaryLayout?: BinaryLayoutService,
webNetworkCapture?: WebNetworkCaptureService,
) => {
const session = createTestBinarySession(() => {
throw new Error("selected views must not start a deep provider");
});
const server = createServer(
{ kind: "session", session },
{
...(binaryLayout === undefined ? {} : { binaryLayout }),
...(webNetworkCapture === undefined ? {} : { webNetworkCapture }),
},
);
const client = new Client({ name: "analysis-view-mcp", version: "1" });
onTestFinished(async () => {
await client.close();
await server.close();
await session.close();
});
const [clientTransport, serverTransport] =
InMemoryTransport.createLinkedPair();
await server.connect(serverTransport);
await client.connect(clientTransport);
return { client, session };
};
it("preserves native relationship coverage through inline and retained MCP views", async () => {
const { client, session } = await connect();
const { ajv, outputSchema } = await advertisedMcpToolSchemas(
client,
"inspect_analysis_view",
);
const coverages: (RelationshipCoverage | undefined)[] = [
{
status: "partial",
limited_directions: ["callees"],
limitations: ["Callees: fixture read limit"],
diagnostics: ["Fixture relationship diagnostic"],
},
{ status: "partial", limitations: ["Historical read limit"] },
{
status: "unknown",
limited_directions: ["callees"],
reason: "Callees not measured",
diagnostics: ["Unknown coverage diagnostic"],
},
{
status: "complete",
diagnostics: ["Only direct references are observed"],
},
{ status: "complete" },
undefined,
];
for (const coverage of coverages) {
const parent = analysisViewNativeEvidence(coverage);
const parentBefore = JSON.stringify(parent);
const recorded = session.recordEvidence(parent);
if (!recorded.ok) throw recorded.error;
const outputs = [];
for (const source of [
{ kind: "inline", evidence: parent },
{ kind: "retained-evidence", evidence_id: parent.evidence_id },
]) {
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source,
view: { kind: "native", facet: "callees", offset: 0, limit: 1 },
},
});
expect(response.isError).not.toBe(true);
expect(
ajv.validate(outputSchema, response.structuredContent),
JSON.stringify(ajv.errors),
).toBe(true);
const evidence = parseEvidence(response.structuredContent);
expect(evidence.relationship_coverage).toEqual(
parent.relationship_coverage,
);
expect(evidence.normalized_result).toMatchObject({
item: [],
coverage: { total: 0, exhausted: true },
limitations: expect.arrayContaining(["Fixture dossier caveat"]),
});
expect(session.evidenceById(evidence.evidence_id)).toEqual(evidence);
outputs.push(evidence);
}
expect(outputs[0]).toEqual(outputs[1]);
expect(JSON.stringify(parent)).toBe(parentBefore);
}
});
it("advertises typed module references and preserves portable/retained Evidence parity", async () => {
const { client, session } = await connect();
const parent = analysisViewJavaScriptEvidence(
analysisViewJavaScriptAnalysisWithSource(),
);
const recorded = session.recordEvidence(parent);
if (!recorded.ok) throw recorded.error;
const { ajv, inputSchema, outputSchema } = await advertisedMcpToolSchemas(
client,
"inspect_analysis_view",
);
const view = {
kind: "javascript",
collection: "modules",
projection: "references",
where: { path: "renderer.js" },
limit: 1,
};
const outputs = [];
for (const source of [
{ kind: "inline", evidence: parent },
{ kind: "retained-evidence", evidence_id: parent.evidence_id },
]) {
expect(ajv.validate(inputSchema, { source, view })).toBe(true);
expect(
ajv.validate(inputSchema, {
source,
view: { ...view, where: { glob: "*" } },
}),
).toBe(false);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: { source, view },
});
expect(response.isError).not.toBe(true);
expect(
ajv.validate(outputSchema, response.structuredContent),
JSON.stringify(ajv.errors),
).toBe(true);
const evidence = parseEvidence(response.structuredContent);
expect(evidence).toMatchObject({
evidence_links: [parent.evidence_id],
normalized_result: {
kind: "javascript",
rows: [
{
path: "renderer.js",
identity: expect.any(Object),
basis: expect.any(Array),
},
],
coverage: { exhausted: true },
},
});
expect(session.evidenceById(evidence.evidence_id)).toEqual(evidence);
outputs.push(evidence.normalized_result);
}
expect(outputs[0]).toEqual(outputs[1]);
});
it("projects opted-in query names identically from retained and portable MCP Evidence", async () => {
const { client, session } = await connect();
const capture = analysisViewCaptureFixture();
capture.records = capture.records.slice(0, 2);
capture.total_records = 2;
capture.records.forEach((record, index) => {
record.reported = {
request: {
method: "GET",
url:
index === 0
? "https://api.test/products?a+b=one&a%2Bb=two&=three&a+b=four"
: "https://api.test/products?a+b=five&snow=six#ignored",
},
};
});
const parent = analysisViewCaptureEvidence(capture);
const retained = session.recordEvidence(parent);
if (!retained.ok) throw retained.error;
const outputs = [];
for (const source of [
{ kind: "inline", evidence: parent },
{ kind: "retained-evidence", evidence_id: parent.evidence_id },
]) {
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source,
view: { kind: "api-surface", include_query_parameter_names: true },
},
});
expect(response.isError).not.toBe(true);
const evidence = parseEvidence(response.structuredContent);
expect(evidence.normalized_result).toMatchObject({
parent_evidence_id: parent.evidence_id,
groups: [
{
query_parameter_names: [
{ name: "", record_ordinals: [0] },
{ name: "a b", record_ordinals: [0, 1] },
{ name: "a+b", record_ordinals: [0] },
{ name: "snow", record_ordinals: [1] },
],
},
],
});
outputs.push(evidence.normalized_result);
}
expect(outputs[0]).toEqual(outputs[1]);
});
it("discovers an oversized native dossier by exact metadata and reads its retained view", async () => {
const { client, session } = await connect();
const advertised = (await client.listTools()).tools.find(
(tool) => tool.name === "inspect_analysis_view",
);
expect(advertised?.description).toContain(
'Native function Evidence requires kind native: view: {kind: "native", facet: "procedure"}',
);
const provider = { id: "ghidra", name: "Ghidra", version: "12.1.4" };
const profile = createAnalysisProfile(provider, { language: "fixture" });
const parent = createEvidence(
{ path: "/fixtures/large.exe", format: "pe", sha256: "a".repeat(64) },
provider,
{
operation: "analyze_function",
parameters: { procedure: "0x401000" },
analysisProfile: profile,
result: {
...jsonObjectSchema.parse(ghidraFunctionDossier()),
pseudocode: "x".repeat(6 * 1024 * 1024),
},
rawResult: null,
},
);
const other = createEvidence(
{ path: "/fixtures/other.exe", format: "pe", sha256: "b".repeat(64) },
provider,
{
operation: "analyze_function",
parameters: {},
result: ghidraFunctionDossier(),
},
);
for (const record of [parent, other]) {
const retained = session.recordEvidence(record);
if (!retained.ok) throw retained.error;
}
const unsupportedSummary = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "retained-evidence", evidence_id: parent.evidence_id },
view: { kind: "summary" },
},
});
expect(parseMcpToolError(unsupportedSummary).error).toMatchObject({
code: "invalid_request",
details: {
issues: [
{
path: ["view"],
message: "analyze_function Evidence requires a native view.",
},
],
},
});
const broad = await client.callTool({
name: "get_evidence_bundle",
arguments: {},
});
const broadError = parseMcpToolError(broad).error;
expect(broadError.code).toBe("resource_constraint");
expect(broadError.remediation?.action).toContain(
"get_evidence_bundle and detail: summary",
);
const response = await client.callTool({
name: "get_evidence_bundle",
arguments: {
detail: "summary",
filters: {
operation: "analyze_function",
target_sha256: "a".repeat(64),
analysis_profile_digest: profile.digest,
procedure_address: "0x401000",
},
},
});
expect(response.isError).not.toBe(true);
const result = toolContract("get_evidence_bundle").outputSchema.parse(
response.structuredContent,
).result;
expect(result).toMatchObject({
kind: "evidence-bundle-summary",
total_retained_records: 2,
matching_records: 1,
records: [
{
evidence_id: parent.evidence_id,
analysis_profile_digest: profile.digest,
retention: "complete-record",
native_dossier: { available: true, procedure_address: "0x401000" },
},
],
});
expect(JSON.stringify(response)).not.toContain("pseudocode");
expect(
JSON.parse(response.content.find((c) => c.type === "text")?.text ?? "null"),
).toEqual(response.structuredContent);
const page = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "retained-evidence", evidence_id: parent.evidence_id },
view: { kind: "native", facet: "pseudocode", offset: 0, limit: 32 },
},
});
expect(page.isError).not.toBe(true);
expect(
toolContract("inspect_analysis_view").outputSchema.parse(
page.structuredContent,
).normalized_result,
).toMatchObject({
item: { text: "x".repeat(32) },
coverage: { total: 6 * 1024 * 1024, exhausted: false },
});
const empty = await client.callTool({
name: "get_evidence_bundle",
arguments: {
detail: "summary",
filters: { evidence_id: `ev_${"e".repeat(64)}` },
},
});
expect(
toolContract("get_evidence_bundle").outputSchema.parse(
empty.structuredContent,
).result,
).toMatchObject({ matching_records: 0, records: [] });
await client.ping();
const invalid = await client.callTool({
name: "get_evidence_bundle",
arguments: { filters: { operation: "analyze_function" } },
});
expect(parseMcpToolError(invalid).error.code).toBe("invalid_request");
});
it("advertises exact schemas and projects one layout section from retained Evidence", async () => {
const layout = analysisViewLayoutFixture();
const service = new BinaryLayoutService({
inspectAndClose(input, options) {
return this.inspect(input, options);
},
identity: BINARY_LAYOUT_TEST_PROVIDER,
inspect: () => Promise.resolve(ok(layout)),
});
const { client, session } = await connect(service);
const { ajv, inputSchema, outputSchema } = await advertisedMcpToolSchemas(
client,
"inspect_analysis_view",
);
expect(
ajv.validate(inputSchema, {
source: { kind: "inline", evidence: analysisViewLayoutEvidence() },
view: { kind: "summary" },
}),
).toBe(true);
expect(
ajv.validate(inputSchema, {
source: { kind: "inline", evidence: {} },
view: { kind: "summary" },
}),
).toBe(false);
expect(
ajv.validate(inputSchema, {
source: {
kind: "retained-evidence",
evidence_id: `ev_${"a".repeat(64)}`,
},
view: { kind: "summary", approval: true },
}),
).toBe(false);
const inspected = await client.callTool({
name: "inspect_binary_layout",
arguments: { path: layout.artifact.path },
});
expect(inspected.isError).not.toBe(true);
const parent = toolContract("inspect_binary_layout").outputSchema.parse(
inspected.structuredContent,
);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: {
kind: "retained-evidence",
evidence_id: parent.evidence_id,
},
view: {
kind: "item",
collection: "sections",
selector: { name: ".data" },
},
},
});
expect(response.isError).not.toBe(true);
expect(
ajv.validate(outputSchema, response.structuredContent),
JSON.stringify(ajv.errors),
).toBe(true);
const parsed = toolContract("inspect_analysis_view").outputSchema.parse(
response.structuredContent,
);
expect(parsed.normalized_result).toMatchObject({
kind: "item",
parent_evidence_id: parent.evidence_id,
item: { name: { display: ".data" } },
});
expect(session.evidenceById(parsed.evidence_id)).toEqual(
parseEvidence(parsed),
);
});
it("points omitted inline messages to the caller's parent without promising session recovery", async () => {
const { client, session } = await connect();
const messages = Array.from(
{ length: 65 },
(_, index) => `Limitation ${String(index).padStart(2, "0")}`,
).concat("z".repeat(300_000));
const parent = analysisViewLayoutEvidence({
...analysisViewLayoutFixture(),
limitations: messages,
});
expect(session.evidenceById(parent.evidence_id)).toBeUndefined();
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "inline", evidence: parent },
view: { kind: "summary" },
},
});
expect(response.isError).not.toBe(true);
const view = toolContract("inspect_analysis_view").outputSchema.parse(
response.structuredContent,
);
const projected = z
.object({ limitations: z.array(z.string()) })
.parse(view.normalized_result);
expect(projected.limitations).toEqual([
...messages.slice(0, 65),
`1 additional limitations omitted from this view; consult source parent Evidence ${parent.evidence_id} for the complete list (the caller-supplied record, or its retaining session).`,
]);
expect(session.evidenceById(parent.evidence_id)).toBeUndefined();
expect(session.evidenceById(view.evidence_id)).toEqual(parseEvidence(view));
});
it("reports a tampered inline Evidence ID at its full MCP source path", async () => {
const { client } = await connect();
const parent = analysisViewLayoutEvidence();
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: {
kind: "inline",
evidence: { ...parent, evidence_id: `ev_${"0".repeat(64)}` },
},
view: { kind: "summary" },
},
});
expect(parseMcpToolError(response).error).toMatchObject({
code: "invalid_request",
category: "invalid_input",
details: {
issues: [
{
path: ["source", "evidence", "evidence_id"],
reason: "invalid_value",
message: "Evidence semantic identifier does not match its record",
},
],
},
});
});
it("rejects self-contained WASM and EVM Evidence views when the subject format differs", async () => {
const { client } = await connect();
const wasm = wasmArtifactFixture();
const evm = evmInterfaceFixture();
const cases = [
{
evidence: createEvidence(
{
path: wasm.artifact.path,
format: "elf",
sha256: wasm.artifact.sha256,
},
{ id: "fixture-wabt", name: "WABT", version: "1" },
{
operation: "inspect_wasm_artifact",
parameters: { path: wasm.artifact.path },
result: wasm,
},
),
},
{
evidence: createEvidence(
{
path: evm.artifact.path,
format: "elf",
sha256: evm.artifact.sha256,
},
{ id: "fixture-evm", name: "EVMole", version: "1" },
{
operation: "inspect_evm_interface",
parameters: { path: evm.artifact.path, encoding: "hex" },
result: evm,
},
),
},
] as const;
for (const { evidence } of cases) {
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "inline", evidence },
view: { kind: "summary" },
},
});
expect(parseMcpToolError(response).error).toMatchObject({
code: "invalid_request",
details: {
issues: [
expect.objectContaining({
path: ["source", "subject"],
reason: "invalid_value",
}),
],
},
});
}
});
it("admits portable captures before SDK Evidence parsing and preserves capture caveats", async () => {
const { client } = await connect();
const dense = analysisViewCaptureEvidence(analysisViewObjectDenseCapture());
const refused = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "inline", evidence: dense },
view: { kind: "api-surface" },
},
});
expect(parseMcpToolError(refused).error).toMatchObject({
code: "resource_constraint",
details: { resource: "memory" },
});
await client.ping();
const capture = analysisViewCaptureFixture();
capture.records = capture.records.slice(0, 1);
capture.total_records = 1;
const caveat = "Producer omitted exchanges before capture startup.";
capture.limitations = [caveat, caveat];
capture.records[0]!.reported = {
request: { method: "GET", url: "https://api.test/a" },
response: { status: 200, content: { mimeType: "\u00a0text/plain\u00a0" } },
};
const parent = analysisViewCaptureEvidence(capture);
expect(parent.limitations).not.toContain(caveat);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "inline", evidence: parent },
view: { kind: "api-surface" },
},
});
const result = parseEvidence(response.structuredContent);
expect(result.limitations.filter((value) => value === caveat)).toEqual([
caveat,
]);
expect(result.normalized_result).toMatchObject({
groups: [{ content_types: [], flags: ["single-sample"] }],
limitations: expect.arrayContaining([caveat]),
unknowns: [
"Some retained records have unavailable status or content type values.",
],
});
});
it("rejects inconsistent portable coverage and accepts caller-ordered ordinal subsets", async () => {
const { client } = await connect();
for (const [total, ordinals] of [
[0, [0]],
[2, [0, 0]],
[2, [2]],
] as const) {
const capture = analysisViewCaptureFixture();
capture.total_records = total;
capture.records = ordinals.map((ordinal) => ({
...capture.records[0]!,
ordinal,
}));
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: {
kind: "inline",
evidence: analysisViewCaptureEvidence(capture),
},
view: { kind: "api-surface" },
},
});
expect(parseMcpToolError(response).error.code).toBe("invalid_request");
}
const capture = analysisViewCaptureFixture();
capture.records = [capture.records[8]!, capture.records[1]!];
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: {
kind: "inline",
evidence: analysisViewCaptureEvidence(capture),
},
view: { kind: "api-surface" },
},
});
expect(
parseEvidence(response.structuredContent).normalized_result,
).toMatchObject({
records: { examined: 2, total: 15 },
limitations: expect.arrayContaining([
"Parent contains an ordinal subset; the surface covers only retained records.",
]),
});
});
it("delivers a bounded native dossier view over MCP without a Ghidra provider or lost connection", async () => {
const { client, session } = await connect();
const original = functionDossierSchema.parse(ghidraFunctionDossier());
const parent = createEvidence(
{
path: "/fixtures/oversized-native.exe",
format: "pe",
sha256: "b".repeat(64),
},
{ id: "ghidra", name: "Ghidra", version: "12.1.4" },
{
operation: "analyze_function",
parameters: { address: "0x401000" },
result: { ...original, pseudocode: "X".repeat(11 * 1024 * 1024) },
},
);
expect(session.recordEvidence(parent).ok).toBe(true);
const advertised = (await client.listTools()).tools.find(
(tool) => tool.name === "inspect_analysis_view",
);
if (advertised?.outputSchema === undefined)
throw new Error("missing view schemas");
const ajv = new Ajv2020({ strict: false, validateFormats: false });
const inputSchema: Record<string, unknown> = advertised.inputSchema;
const outputSchema: Record<string, unknown> = advertised.outputSchema;
const arguments_ = {
source: { kind: "retained-evidence", evidence_id: parent.evidence_id },
view: { kind: "native", facet: "pseudocode", offset: 0, limit: 500 },
};
expect(ajv.validateSchema(inputSchema)).toBe(true);
expect(ajv.validateSchema(outputSchema)).toBe(true);
expect(
ajv.validate(inputSchema, arguments_),
JSON.stringify(ajv.errors),
).toBe(true);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: arguments_,
});
expect(response.isError).not.toBe(true);
expect(
ajv.validate(outputSchema, response.structuredContent),
JSON.stringify(ajv.errors),
).toBe(true);
const parsed = toolContract("inspect_analysis_view").outputSchema.parse(
response.structuredContent,
);
expect(parsed.normalized_result).toMatchObject({
kind: "native",
parent_evidence_id: parent.evidence_id,
procedure_address: "0x401000",
coverage: { examined: 500, next_offset: 500, total: 11 * 1024 * 1024 },
});
expect(parsed.evidence_links).toEqual([parent.evidence_id]);
expect(parsed.locations).toEqual([
{ kind: "artifact-path", path: "/fixtures/oversized-native.exe" },
{ kind: "address", address: "0x401000" },
]);
expect(JSON.stringify(response.structuredContent).length).toBeLessThan(12000);
await client.ping();
});
it("retains an oversized selected observation and keeps subsequent views usable", async () => {
const { client, session } = await connect();
const analysis = analysisViewJavaScriptAnalysisWithSource();
const original = analysis.graph.nodes[0];
if (original === undefined) throw new Error("missing module");
const source = "\0".repeat(1_000_000);
const node = createJavaScriptApplicationNode({
kind: original.kind,
identity: original.identity,
observations: original.observations.map((observation) => ({
label: observation.label,
properties: { ...observation.properties, source },
evidence: observation.evidence,
})),
});
const graph = createJavaScriptApplicationGraph({
schema: "JavaScriptApplicationGraph",
root_node_ids: [node.node_id],
nodes: [node],
edges: [],
coverage: analysis.graph.coverage,
limitations: analysis.graph.limitations,
});
const parent = analysisViewJavaScriptEvidence(
analysisViewBindJavaScriptGraphs(analysis, graph),
);
expect(session.recordEvidence(parent).ok).toBe(true);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "retained-evidence", evidence_id: parent.evidence_id },
view: {
kind: "item",
collection: "modules",
selector: { node_id: node.node_id },
},
},
});
expect(response.isError).toBe(true);
const reference = z
.object({
error: z.object({
details: z.object({
reported_limits: z.object({
evidence_reference: z.object({
kind: z.literal("retained-evidence"),
evidence_id: z.string(),
}),
}),
}),
}),
})
.parse(parseMcpToolError(response)).error.details
.reported_limits.evidence_reference;
expect(
session.evidenceById(reference.evidence_id)?.normalized_result,
).toMatchObject({
item: {
observations: [
expect.objectContaining({
properties: expect.objectContaining({ source }),
}),
],
},
});
await client.ping();
const summary = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "retained-evidence", evidence_id: parent.evidence_id },
view: { kind: "summary" },
},
});
expect(summary.isError).not.toBe(true);
});
it("retains an oversized native assembly row and permits a subsequent smaller view", async () => {
const { client, session } = await connect();
const row = "\0".repeat(1_000_000);
const parent = createEvidence(
{ path: "/fixtures/large-row.exe", format: "pe", sha256: "c".repeat(64) },
{ id: "ghidra", name: "Ghidra", version: "12.1.4" },
{
operation: "analyze_function",
parameters: {},
result: {
...functionDossierSchema.parse(ghidraFunctionDossier()),
assembly: [row],
},
},
);
expect(session.recordEvidence(parent).ok).toBe(true);
const source = { kind: "retained-evidence", evidence_id: parent.evidence_id };
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source,
view: { kind: "native", facet: "assembly", offset: 0, limit: 1 },
},
});
expect(response.isError).toBe(true);
const error = z
.object({
error: z.object({
details: z.object({
reported_limits: z.object({
evidence_reference: z.object({
kind: z.literal("retained-evidence"),
evidence_id: z.string(),
}),
}),
}),
}),
})
.parse(parseMcpToolError(response));
const retained = session.evidenceById(
error.error.details.reported_limits.evidence_reference.evidence_id,
);
expect(retained?.normalized_result).toMatchObject({
item: [row],
parent_evidence_id: parent.evidence_id,
});
const next = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source,
view: { kind: "native", facet: "pseudocode", offset: 0, limit: 12 },
},
});
expect(next.isError).not.toBe(true);
expect(
toolContract("inspect_analysis_view").outputSchema.parse(
next.structuredContent,
).normalized_result,
).toMatchObject({
item: { text: "int fixture_" },
parent_evidence_id: parent.evidence_id,
});
await client.ping();
});
it("reads successive native source pages directly from a completed-result transport error", async () => {
const pseudocode = "int fixture_add(int a, int b) { return a + b; }\n".repeat(
150_000,
);
const harness = await connectGhidraMcp(
"native-continuation",
async (operation) => {
if (operation !== "analyze_function")
throw new Error(`Unexpected provider operation: ${operation}`);
return ok({
value: {
...functionDossierSchema.parse(ghidraFunctionDossier()),
pseudocode,
},
limitations: [],
});
},
);
onTestFinished(() => harness.close());
const { mcp: client, session, calls } = harness;
const failed = await client.callTool({
name: "analyze_function",
arguments: { procedure: "0x401000" },
});
expect(failed.isError).toBe(true);
const limits = z
.object({
error: z.object({
details: z.object({ reported_limits: jsonObjectSchema }),
}),
})
.parse(parseMcpToolError(failed)).error.details.reported_limits;
const reference = z
.object({
kind: z.literal("retained-evidence"),
evidence_id: z.string(),
})
.parse(limits.evidence_reference);
const parent = parseEvidence(session.evidenceById(reference.evidence_id));
expect(parent.operation).toBe("analyze_function");
expect(parent.parameters).toMatchObject({ procedure: "0x401000" });
const call = z
.object({
name: z.literal("inspect_analysis_view"),
arguments: toolContract("inspect_analysis_view").inputSchema,
})
.parse(limits.continuation);
expect(call.arguments.source).toEqual({
kind: "retained-evidence",
evidence_id: parent.evidence_id,
});
const first = await client.callTool(call);
expect(first.isError).not.toBe(true);
const page = toolContract("inspect_analysis_view").outputSchema.parse(
first.structuredContent,
).normalized_result;
expect(page).toMatchObject({
parent_evidence_id: parent.evidence_id,
item: { text: pseudocode.slice(0, 64) },
coverage: { examined: 64, next_offset: 64, exhausted: false },
});
expect(Buffer.byteLength(JSON.stringify(first))).toBeLessThan(
STDIO_DEFAULT_MAX_BUFFER_SIZE,
);
const second = await client.callTool({
name: call.name,
arguments: {
...call.arguments,
view: {
kind: "native",
facet: "pseudocode",
offset: page.coverage.next_offset,
limit: 64,
},
},
});
expect(second.isError).not.toBe(true);
expect(
toolContract("inspect_analysis_view").outputSchema.parse(
second.structuredContent,
).normalized_result,
).toMatchObject({ item: { text: pseudocode.slice(64, 128) } });
expect(
session.evidenceById(parent.evidence_id)?.normalized_result,
).toMatchObject({ pseudocode });
await client.ping();
expect(calls).toEqual(["analyze_function"]);
});
it("returns a JavaScript summary without graph payloads from a retained reference", async () => {
const { client, session } = await connect();
const parent = analysisViewJavaScriptEvidence();
expect(session.recordEvidence(parent).ok).toBe(true);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: {
kind: "retained-evidence",
evidence_id: parent.evidence_id,
},
view: { kind: "summary" },
},
});
expect(response.isError).not.toBe(true);
const parsed = toolContract("inspect_analysis_view").outputSchema.parse(
response.structuredContent,
);
expect(parsed.normalized_result).toMatchObject({
kind: "summary",
parent_operation: "analyze_javascript_application",
summary: { format: "directory" },
});
expect(parsed.normalized_result).not.toHaveProperty("summary.semantic_graph");
const stale = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: {
kind: "retained-evidence",
evidence_id: `ev_${"e".repeat(64)}`,
},
view: { kind: "summary" },
},
});
expect(parseMcpToolError(stale)).toMatchObject({
error: { details: { reason: "missing" } },
});
});
it("accepts a transport-constraint retained reference as the view source", async () => {
const { client, session } = await connect();
const parent = analysisViewJavaScriptEvidence();
expect(session.recordEvidence(parent).ok).toBe(true);
const evidenceReference = {
kind: "retained-evidence" as const,
evidence_id: parent.evidence_id,
};
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: evidenceReference,
view: {
kind: "page",
collection: "modules",
offset: 0,
limit: 8,
},
},
});
expect(response.isError).not.toBe(true);
const parsed = toolContract("inspect_analysis_view").outputSchema.parse(
response.structuredContent,
);
expect(parsed.normalized_result).toMatchObject({
kind: "page",
parent_evidence_id: parent.evidence_id,
coverage: { exhausted: true },
});
if (parsed.normalized_result.kind !== "page")
throw new Error("expected page view");
expect(parsed.normalized_result.items).toEqual([
{
node_id: expect.any(String),
kind: "javascript-asset",
path: "renderer.js",
},
]);
});
it("projects an actual HAR capture retained by the public MCP workflow", async () => {
const root = await createTestTempDirectory("rea-surface-mcp-");
const path = join(root, "surface.har");
await writeFile(path, JSON.stringify(apiSurfaceHar()));
const service = new WebNetworkCaptureService({
inspect: async (input) => {
const bytes = await readFile(input.capture_path);
return ok({
...decodeHarCapture(bytes.toString("utf8"), input.sensitive_values),
artifact: {
path: input.capture_path,
sha256: createHash("sha256").update(bytes).digest("hex"),
bytes: bytes.length,
},
format: "har",
runtime_attribution: "unknown",
limitations: [],
});
},
});
const { client } = await connect(undefined, service);
const captureResponse = await client.callTool({
name: "inspect_web_network_capture",
arguments: { capture_path: path, format: "har" },
});
expect(captureResponse.isError).not.toBe(true);
const parent = parseEvidence(captureResponse.structuredContent);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "retained-evidence", evidence_id: parent.evidence_id },
view: { kind: "api-surface" },
},
});
expect(response.isError).not.toBe(true);
const evidence = parseEvidence(response.structuredContent);
expect(evidence.confidence).toBe("inferred");
expect(evidence.normalized_result).toMatchObject({
records: { examined: 13, total: 13, excluded: 0 },
coverage: { total: 12, exhausted: true },
groups: expect.arrayContaining([
expect.objectContaining({
path_template: "/users/{numeric}",
record_ordinals: [0, 1],
}),
]),
});
});
const PUBLISHED_EXAMPLES: readonly (readonly [
name: string,
example: {
readonly input: {
readonly source: { readonly evidence_id: string };
readonly view: Record<string, unknown>;
};
readonly output: { readonly artifact: { readonly path: string } };
},
fixture: () => unknown,
])[] = [
[
"API surface",
API_SURFACE_EXAMPLE,
() => {
const fixture = apiSurfaceHar();
fixture.log.entries = fixture.log.entries.slice(0, 2);
return fixture;
},
],
[
"environment manifest",
ENVIRONMENT_MANIFEST_EXAMPLE,
() => {
const fixture = environmentManifestHar();
fixture.log.entries = [10, 20, 27].map(
(index) => fixture.log.entries[index]!,
);
return fixture;
},
],
];
it.each(PUBLISHED_EXAMPLES)(
"reproduces the published %s example from decoded fixture bytes",
async (_name, example, fixture) => {
const bytes = Buffer.from(JSON.stringify(fixture()));
const path = example.output.artifact.path;
const service = new WebNetworkCaptureService({
inspect: async () =>
ok({
...decodeHarCapture(bytes.toString("utf8"), []),
artifact: {
path,
sha256: createHash("sha256").update(bytes).digest("hex"),
bytes: bytes.length,
},
format: "har",
runtime_attribution: "unknown",
limitations: [],
}),
});
const { client } = await connect(undefined, service);
const capture = await client.callTool({
name: "inspect_web_network_capture",
arguments: { capture_path: path, format: "har" },
});
expect(parseEvidence(capture.structuredContent).evidence_id).toBe(
example.input.source.evidence_id,
);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: example.input,
});
expect(parseEvidence(response.structuredContent).normalized_result).toEqual(
example.output,
);
},
);
it("extracts an environment state identically from retained, portable and file MCP Evidence", async () => {
const { client, session } = await connect();
const { ajv, inputSchema, outputSchema } = await advertisedMcpToolSchemas(
client,
"inspect_analysis_view",
);
const capture = environmentStateCaptureFixture();
const parent = analysisViewCaptureEvidence(capture);
expect(session.recordEvidence(parent).ok).toBe(true);
const evidencePath = join(
await createTestTempDirectory("rea-state-file-mcp-"),
"capture.json",
);
await writeFile(evidencePath, JSON.stringify(parent));
const view = {
kind: "environment-state",
template_rules: ["after-uid-literal", "opaque-token", "sibling-variance"],
};
const outputs = [];
for (const source of [
{ kind: "inline", evidence: parent },
{ kind: "retained-evidence", evidence_id: parent.evidence_id },
{ kind: "evidence-file", path: evidencePath },
]) {
expect(ajv.validate(inputSchema, { source, view })).toBe(true);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: { source, view },
});
expect(response.isError).not.toBe(true);
expect(
ajv.validate(outputSchema, response.structuredContent),
JSON.stringify(ajv.errors),
).toBe(true);
const evidence = parseEvidence(response.structuredContent);
expect(evidence.confidence).toBe("inferred");
expect(evidence.normalized_result).toMatchObject({
kind: "environment-state",
views: expect.arrayContaining([
expect.objectContaining({
tool: "get_folders_by_uid",
shape: "item",
}),
]),
});
expect(session.evidenceById(evidence.evidence_id)).toEqual(evidence);
outputs.push(evidence.normalized_result);
}
expect(outputs[0]).toEqual(outputs[1]);
expect(outputs[2]).toEqual(outputs[0]);
const hidden = createEvidence(
{
path: capture.artifact.path,
sha256: capture.artifact.sha256!,
format: "file",
},
{ id: "fixture", name: "fixture", version: "1" },
{
operation: "inspect_web_network_capture",
parameters: { sensitive_value_count: 1 },
result: capture,
confidence: "observed",
},
);
const refused = await client.callTool({
name: "inspect_analysis_view",
arguments: { source: { kind: "inline", evidence: hidden }, view },
});
expect(parseMcpToolError(refused).error.code).toBe("unsupported_target");
expect(JSON.stringify(refused)).not.toContain("probe folder");
await client.ping();
});
it("projects an environment manifest identically from retained, portable and file MCP Evidence", async () => {
const { client, session } = await connect();
const { ajv, inputSchema, outputSchema } = await advertisedMcpToolSchemas(
client,
"inspect_analysis_view",
);
const capture = environmentManifestCaptureFixture();
const parent = analysisViewCaptureEvidence(capture);
expect(session.recordEvidence(parent).ok).toBe(true);
const evidencePath = join(
await createTestTempDirectory("rea-evidence-file-mcp-"),
"capture.json",
);
await writeFile(evidencePath, JSON.stringify(parent));
const view = {
kind: "environment-manifest",
template_rules: ["after-uid-literal", "opaque-token", "sibling-variance"],
};
const outputs = [];
for (const source of [
{ kind: "inline", evidence: parent },
{ kind: "retained-evidence", evidence_id: parent.evidence_id },
{ kind: "evidence-file", path: evidencePath },
]) {
expect(ajv.validate(inputSchema, { source, view })).toBe(true);
expect(
ajv.validate(inputSchema, {
source,
view: { ...view, template_rules: ["glob"] },
}),
).toBe(false);
const response = await client.callTool({
name: "inspect_analysis_view",
arguments: { source, view },
});
expect(response.isError).not.toBe(true);
expect(
ajv.validate(outputSchema, response.structuredContent),
JSON.stringify(ajv.errors),
).toBe(true);
const evidence = parseEvidence(response.structuredContent);
expect(evidence.confidence).toBe("inferred");
expect(evidence.normalized_result).toMatchObject({
kind: "environment-manifest",
name: "grafana-example",
templated_groups: 13,
tools: expect.arrayContaining([
expect.objectContaining({
name: "get_plugins_settings_by_plugin_id",
path_template: "/api/plugins/{plugin_id}/settings",
}),
expect.objectContaining({
name: "get_folders_by_uid",
path_template: "/api/folders/{uid}",
}),
]),
});
expect(session.evidenceById(evidence.evidence_id)).toEqual(evidence);
for (const value of ["probe folder", "Main Org", "JFIF", "console.log"])
expect(JSON.stringify(response)).not.toContain(value);
outputs.push(evidence.normalized_result);
}
expect(outputs[0]).toEqual(outputs[1]);
expect(outputs[2]).toEqual(outputs[0]);
// JSON Schema cannot express "absolute"; the tool refuses a relative path when called.
const relative = await client.callTool({
name: "inspect_analysis_view",
arguments: {
source: { kind: "evidence-file", path: "capture.json" },
view,
},
});
expect(parseMcpToolError(relative).error.code).toBe("invalid_request");
await writeFile(evidencePath, JSON.stringify({ hello: 1 }));
const invalidFile = await client.callTool({
name: "inspect_analysis_view",
arguments: { source: { kind: "evidence-file", path: evidencePath }, view },
});
expect(parseMcpToolError(invalidFile).error).toMatchObject({
details: {
path: evidencePath,
reason: "wrong-kind",
content_issues: expect.arrayContaining([
{
path: ["evidence_id"],
reason: "missing_argument",
expected: "string",
},
]),
},
});
const hidden = createEvidence(
{
path: capture.artifact.path,
sha256: capture.artifact.sha256!,
format: "file",
},
{ id: "fixture", name: "fixture", version: "1" },
{
operation: "inspect_web_network_capture",
parameters: { sensitive_value_count: 1 },
result: capture,
confidence: "observed",
},
);
const refused = await client.callTool({
name: "inspect_analysis_view",
arguments: { source: { kind: "inline", evidence: hidden }, view },
});
expect(parseMcpToolError(refused).error.code).toBe("unsupported_target");
expect(JSON.stringify(refused)).toContain("sensitive_values_unretained");
expect(JSON.stringify(refused)).not.toContain("/api/folders");
await client.ping();
});