feat(agents): add AgentArtifactIndex — join subagents to files they produce

Subscribes to subagentWatcher's subagent:tool_call events and attributes every
Write/Edit/MultiEdit/NotebookEdit file_path to its agent. Dedupes repeat writes,
classifies by extension (html/image/pdf/document/markdown/text/code/other), bounds
memory (FIFO per-agent + per-index caps), and exposes query-by-agent/session/all.

Metadata only — records paths, never reads/serves contents; guard enforcement stays
in the serving layer. recordExternalArtifact() reserved for image-watcher correlation
of Bash-written screenshots. Load-bearing piece for the Agent Gallery view.

Verified against live w7 transcripts: 8 mockups correctly attributed with write counts.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
arkon
2026-06-13 06:07:03 +02:00
co-authored by Claude Opus 4.8
parent 3f2c23cb0f
commit dae4d8fbee
3 changed files with 505 additions and 0 deletions
+292
View File
@@ -0,0 +1,292 @@
/**
* @fileoverview Agent Artifact Index — joins each subagent to the files it produced.
*
* The load-bearing piece behind the Agent Gallery view. It answers "which files did
* this subagent write?" by subscribing to {@link subagentWatcher} events rather than
* re-reading transcripts: every `subagent:tool_call` for a file-producing tool
* (Write/Edit/MultiEdit/NotebookEdit) carries the agent id, session id, and the full
* tool input — including `file_path` — so the join is exact and free of duplicate I/O.
*
* What it stores is METADATA ONLY (paths, kinds, write counts). It never reads or
* serves file contents — guard enforcement (realpath, sensitive-path, attachment-guard,
* extension allowlist) lives in the serving layer that turns these paths into
* attachment/artifact ids. Treat the recorded path as untrusted: it originates from a
* subagent transcript and must pass the guard chain before anything reads it.
*
* Key exports:
* - `AgentArtifactIndex` class — EventEmitter; `start(source?)` / `stop()` lifecycle
* - `agentArtifactIndex` — pre-instantiated singleton
* - `AgentArtifact`, `ArtifactKind`, `ArtifactSource`, `AgentArtifactIndexEvents` — types
*
* Extension point: `recordExternalArtifact()` lets a later image-watcher correlation
* attribute Bash-written screenshots (which never appear as a Write tool_use) to the
* agent that was active when the file landed. The transcript join below stands alone.
*
* @emits artifact:added (new path for an agent), artifact:updated (re-write of a known
* path), artifact:error
* @dependencies subagent-watcher (event source), config/map-limits (bounds)
* @consumedby web/server (SSE broadcast as `agent:artifact`), gallery route
* @module agent-artifact-index
*/
import { EventEmitter } from 'node:events';
import { basename, extname, isAbsolute, resolve } from 'node:path';
import { MAX_ARTIFACTS_PER_AGENT, MAX_TRACKED_AGENTS } from './config/map-limits.js';
import { subagentWatcher, type SubagentToolCall } from './subagent-watcher.js';
// ========== Types ==========
export type ArtifactKind = 'html' | 'image' | 'pdf' | 'document' | 'markdown' | 'text' | 'code' | 'other';
/** How the artifact was attributed to its agent. */
export type ArtifactSource = 'write' | 'edit' | 'notebook' | 'external';
export interface AgentArtifact {
agentId: string;
sessionId: string;
/** Absolute path, normalized (`resolve`d) when absolute; left raw otherwise. */
path: string;
/** `basename(path)`. */
filename: string;
/** Lowercased extension including the dot (e.g. `.html`); `''` when none. */
ext: string;
/** Render hint derived from {@link ext}. Drives the gallery's preview mode. */
kind: ArtifactKind;
source: ArtifactSource;
/** Epoch ms when this path was first attributed to the agent. */
firstSeenAt: number;
/** Epoch ms of the most recent write/edit of this path. */
lastWriteAt: number;
/** Number of write/edit calls observed against this path. */
writeCount: number;
}
/** Typed event map (Node 22 typed EventEmitter, matching {@link BashToolParser}). */
export interface AgentArtifactIndexEvents {
'artifact:added': [AgentArtifact];
'artifact:updated': [AgentArtifact];
'artifact:error': [Error];
}
/** Minimal structural type for the event source, for testability without the real watcher. */
type ToolCallEmitter = Pick<EventEmitter, 'on' | 'off'>;
// ========== Constants ==========
/** Tools whose successful call means the agent produced/changed a file. */
const FILE_WRITE_TOOLS = new Set(['Write', 'Edit', 'MultiEdit', 'NotebookEdit']);
const KIND_BY_EXT: Readonly<Record<string, ArtifactKind>> = {
'.html': 'html',
'.htm': 'html',
'.png': 'image',
'.jpg': 'image',
'.jpeg': 'image',
'.gif': 'image',
'.webp': 'image',
'.avif': 'image',
'.svg': 'image',
'.pdf': 'pdf',
'.docx': 'document',
'.pptx': 'document',
'.xlsx': 'document',
'.md': 'markdown',
'.markdown': 'markdown',
'.txt': 'text',
'.log': 'text',
'.csv': 'text',
'.js': 'code',
'.mjs': 'code',
'.cjs': 'code',
'.ts': 'code',
'.tsx': 'code',
'.jsx': 'code',
'.css': 'code',
'.json': 'code',
'.py': 'code',
'.sh': 'code',
'.yml': 'code',
'.yaml': 'code',
};
// ========== Helpers ==========
function classifyKind(ext: string): ArtifactKind {
return KIND_BY_EXT[ext] ?? 'other';
}
function sourceForTool(tool: string): ArtifactSource {
if (tool === 'Write') return 'write';
if (tool === 'NotebookEdit') return 'notebook';
return 'edit'; // Edit, MultiEdit
}
/**
* Resolve `..`/`.` for absolute paths so repeat writes dedupe to one artifact.
* Relative paths are left raw: `resolve()` would anchor them to the SERVER cwd, not the
* agent's, producing a wrong absolute path. Write requires absolute paths, so this is rare.
*/
function normalizePath(p: string): string {
return isAbsolute(p) ? resolve(p) : p;
}
/** ISO transcript timestamp → epoch ms, falling back to now. */
function timestampToMs(ts: string | undefined): number {
if (ts) {
const n = Date.parse(ts);
if (!Number.isNaN(n)) return n;
}
return Date.now();
}
/** Extract the target path from a file-tool's input (Write/Edit use `file_path`, NotebookEdit `notebook_path`). */
function extractFilePath(input: Record<string, unknown> | undefined): string | undefined {
const candidate = input?.file_path ?? input?.notebook_path;
return typeof candidate === 'string' && candidate.length > 0 ? candidate : undefined;
}
// ========== AgentArtifactIndex ==========
export class AgentArtifactIndex extends EventEmitter<AgentArtifactIndexEvents> {
/** agentId → (normalized path → artifact). Outer map is insertion-ordered for FIFO eviction. */
private readonly byAgent = new Map<string, Map<string, AgentArtifact>>();
private source: ToolCallEmitter | null = null;
private readonly onToolCall = (call: SubagentToolCall): void => {
try {
if (!FILE_WRITE_TOOLS.has(call.tool)) return;
const rawPath = extractFilePath(call.fullInput);
if (!rawPath) return;
this.record(call.agentId, call.sessionId, rawPath, sourceForTool(call.tool), timestampToMs(call.timestamp));
} catch (err) {
this.emit('artifact:error', err instanceof Error ? err : new Error(String(err)));
}
};
/**
* Begin attributing artifacts. Idempotent. Defaults to the real {@link subagentWatcher};
* tests pass a bare EventEmitter and drive `subagent:tool_call` directly.
*/
start(source: ToolCallEmitter = subagentWatcher): void {
if (this.source) {
if (this.source === source) return;
this.stop();
}
this.source = source;
source.on('subagent:tool_call', this.onToolCall as (...args: unknown[]) => void);
}
/** Detach from the event source. Retains recorded artifacts (use {@link clear} to drop them). */
stop(): void {
if (this.source) {
this.source.off('subagent:tool_call', this.onToolCall as (...args: unknown[]) => void);
this.source = null;
}
}
/**
* Attribute a file discovered outside the transcript — e.g. a screenshot written by a
* Bash `capture.mjs`, correlated by the image-watcher to the agent active when it landed.
* Marked `source: 'external'`. (Wired for a follow-up; not yet called by the image-watcher.)
*/
recordExternalArtifact(agentId: string, sessionId: string, absPath: string, atMs: number = Date.now()): void {
if (typeof absPath !== 'string' || absPath.length === 0) return;
this.record(agentId, sessionId, absPath, 'external', atMs);
}
private record(agentId: string, sessionId: string, rawPath: string, source: ArtifactSource, atMs: number): void {
const path = normalizePath(rawPath);
let bucket = this.byAgent.get(agentId);
if (!bucket) {
// Bound the number of tracked agents (FIFO eviction of the oldest agent).
if (this.byAgent.size >= MAX_TRACKED_AGENTS) {
const oldest = this.byAgent.keys().next().value;
if (oldest !== undefined) this.byAgent.delete(oldest);
}
bucket = new Map();
this.byAgent.set(agentId, bucket);
}
const existing = bucket.get(path);
if (existing) {
existing.lastWriteAt = Math.max(existing.lastWriteAt, atMs);
existing.firstSeenAt = Math.min(existing.firstSeenAt, atMs);
existing.writeCount += 1;
existing.sessionId = sessionId; // tolerate late/corrected session attribution
this.emit('artifact:updated', existing);
return;
}
// Bound artifacts per agent (FIFO eviction of the oldest path for this agent).
if (bucket.size >= MAX_ARTIFACTS_PER_AGENT) {
const oldest = bucket.keys().next().value;
if (oldest !== undefined) bucket.delete(oldest);
}
const ext = extname(path).toLowerCase();
const artifact: AgentArtifact = {
agentId,
sessionId,
path,
filename: basename(path),
ext,
kind: classifyKind(ext),
source,
firstSeenAt: atMs,
lastWriteAt: atMs,
writeCount: 1,
};
bucket.set(path, artifact);
this.emit('artifact:added', artifact);
}
/** Artifacts written by one agent, most-recently-written first. */
getArtifactsForAgent(agentId: string): AgentArtifact[] {
const bucket = this.byAgent.get(agentId);
if (!bucket) return [];
return [...bucket.values()].sort((a, b) => b.lastWriteAt - a.lastWriteAt);
}
/** Artifacts across every agent in a Codeman session, most-recently-written first. */
getArtifactsForSession(sessionId: string): AgentArtifact[] {
const out: AgentArtifact[] = [];
for (const bucket of this.byAgent.values()) {
for (const artifact of bucket.values()) {
if (artifact.sessionId === sessionId) out.push(artifact);
}
}
return out.sort((a, b) => b.lastWriteAt - a.lastWriteAt);
}
/** Every tracked artifact, most-recently-written first. */
getAllArtifacts(): AgentArtifact[] {
const out: AgentArtifact[] = [];
for (const bucket of this.byAgent.values()) out.push(...bucket.values());
return out.sort((a, b) => b.lastWriteAt - a.lastWriteAt);
}
getArtifact(agentId: string, path: string): AgentArtifact | undefined {
return this.byAgent.get(agentId)?.get(normalizePath(path));
}
/** Total number of tracked artifacts across all agents. */
count(): number {
let n = 0;
for (const bucket of this.byAgent.values()) n += bucket.size;
return n;
}
/** Drop all artifacts for one agent (e.g. when its subagent window is dismissed). */
clearAgent(agentId: string): void {
this.byAgent.delete(agentId);
}
/** Drop everything. */
clear(): void {
this.byAgent.clear();
}
}
/** Pre-instantiated singleton, mirroring `subagentWatcher`. Call `.start()` to attach. */
export const agentArtifactIndex = new AgentArtifactIndex();
+6
View File
@@ -62,6 +62,12 @@ export const MAX_TRACKED_AGENTS = 500;
*/
export const MAX_PENDING_TOOL_CALLS = 100;
/**
* Maximum distinct file artifacts tracked per subagent in the artifact index.
* Bounds memory for agents that write many files (FIFO eviction when exceeded).
*/
export const MAX_ARTIFACTS_PER_AGENT = 200;
/**
* TTL for orphaned pending tool calls (5 minutes).
* If no tool_result received, entry is cleaned up.
+207
View File
@@ -0,0 +1,207 @@
/**
* Tests for AgentArtifactIndex — the subagent → produced-files join.
*
* Driven through a bare EventEmitter standing in for `subagentWatcher`, so no real
* transcript files, tmux, or server are involved.
*/
import { EventEmitter } from 'node:events';
import { describe, it, expect, beforeEach } from 'vitest';
import { AgentArtifactIndex, type AgentArtifact } from '../src/agent-artifact-index.js';
import { MAX_ARTIFACTS_PER_AGENT } from '../src/config/map-limits.js';
/** Build a `subagent:tool_call` payload like the watcher emits. */
function toolCall(over: {
agentId?: string;
sessionId?: string;
tool: string;
input: Record<string, unknown>;
timestamp?: string;
}) {
return {
agentId: over.agentId ?? 'agent-1',
sessionId: over.sessionId ?? 'sess-1',
timestamp: over.timestamp ?? '2026-06-13T05:44:00.000Z',
tool: over.tool,
input: over.input,
toolUseId: 'toolu_x',
fullInput: over.input,
};
}
describe('AgentArtifactIndex', () => {
let index: AgentArtifactIndex;
let source: EventEmitter;
beforeEach(() => {
source = new EventEmitter();
index = new AgentArtifactIndex();
index.start(source);
});
function emitCall(over: Parameters<typeof toolCall>[0]) {
source.emit('subagent:tool_call', toolCall(over));
}
it('records a Write as an artifact and classifies HTML', () => {
const added: AgentArtifact[] = [];
index.on('artifact:added', (a) => added.push(a));
emitCall({
tool: 'Write',
input: { file_path: '/home/arkon/proj/mockups/01-aurora.html', content: '<!doctype html>' },
});
expect(added).toHaveLength(1);
expect(added[0]).toMatchObject({
agentId: 'agent-1',
sessionId: 'sess-1',
path: '/home/arkon/proj/mockups/01-aurora.html',
filename: '01-aurora.html',
ext: '.html',
kind: 'html',
source: 'write',
writeCount: 1,
});
expect(index.getArtifactsForAgent('agent-1')).toHaveLength(1);
});
it('dedupes repeated writes/edits of the same path and counts them', () => {
const updated: AgentArtifact[] = [];
index.on('artifact:updated', (a) => updated.push(a));
emitCall({ tool: 'Write', input: { file_path: '/p/a.html', content: '1' }, timestamp: '2026-06-13T05:00:00.000Z' });
emitCall({
tool: 'Edit',
input: { file_path: '/p/a.html', old_string: '1', new_string: '2' },
timestamp: '2026-06-13T05:01:00.000Z',
});
emitCall({ tool: 'Edit', input: { file_path: '/p/a.html' }, timestamp: '2026-06-13T05:02:00.000Z' });
const arts = index.getArtifactsForAgent('agent-1');
expect(arts).toHaveLength(1);
expect(arts[0].writeCount).toBe(3);
expect(arts[0].source).toBe('write'); // first attribution wins
expect(arts[0].lastWriteAt).toBe(Date.parse('2026-06-13T05:02:00.000Z'));
expect(arts[0].firstSeenAt).toBe(Date.parse('2026-06-13T05:00:00.000Z'));
expect(updated).toHaveLength(2);
});
it('normalizes ../. in absolute paths so they dedupe', () => {
emitCall({ tool: 'Write', input: { file_path: '/p/sub/../a.html', content: '' } });
emitCall({ tool: 'Write', input: { file_path: '/p/a.html', content: '' } });
expect(index.getArtifactsForAgent('agent-1')).toHaveLength(1);
expect(index.getArtifactsForAgent('agent-1')[0].path).toBe('/p/a.html');
});
it('ignores non-file tools and writes missing a path', () => {
emitCall({ tool: 'Read', input: { file_path: '/p/a.html' } });
emitCall({ tool: 'Bash', input: { command: 'node capture.mjs' } });
emitCall({ tool: 'Write', input: { content: 'no path' } });
expect(index.count()).toBe(0);
});
it('handles NotebookEdit via notebook_path with notebook source', () => {
emitCall({ tool: 'NotebookEdit', input: { notebook_path: '/p/analysis.ipynb', new_source: 'x' } });
const arts = index.getArtifactsForAgent('agent-1');
expect(arts).toHaveLength(1);
expect(arts[0].source).toBe('notebook');
expect(arts[0].kind).toBe('other'); // .ipynb not in the render map
});
it.each([
['/p/x.png', 'image'],
['/p/x.PNG', 'image'],
['/p/x.svg', 'image'],
['/p/x.pdf', 'pdf'],
['/p/x.docx', 'document'],
['/p/x.pptx', 'document'],
['/p/x.md', 'markdown'],
['/p/x.txt', 'text'],
['/p/x.css', 'code'],
['/p/x.ts', 'code'],
['/p/x.bin', 'other'],
['/p/Makefile', 'other'],
])('classifies %s as %s', (file, kind) => {
emitCall({ tool: 'Write', input: { file_path: file, content: '' } });
expect(index.getAllArtifacts()[0].kind).toBe(kind);
});
it('groups artifacts by agent and by session', () => {
emitCall({ agentId: 'a1', sessionId: 's1', tool: 'Write', input: { file_path: '/p/1.html', content: '' } });
emitCall({ agentId: 'a2', sessionId: 's1', tool: 'Write', input: { file_path: '/p/2.html', content: '' } });
emitCall({ agentId: 'a3', sessionId: 's2', tool: 'Write', input: { file_path: '/p/3.html', content: '' } });
expect(index.getArtifactsForAgent('a1')).toHaveLength(1);
expect(
index
.getArtifactsForSession('s1')
.map((a) => a.agentId)
.sort()
).toEqual(['a1', 'a2']);
expect(index.getArtifactsForSession('s2')).toHaveLength(1);
expect(index.count()).toBe(3);
});
it('sorts results most-recently-written first', () => {
emitCall({
tool: 'Write',
input: { file_path: '/p/old.html', content: '' },
timestamp: '2026-06-13T01:00:00.000Z',
});
emitCall({
tool: 'Write',
input: { file_path: '/p/new.html', content: '' },
timestamp: '2026-06-13T09:00:00.000Z',
});
expect(index.getArtifactsForAgent('agent-1').map((a) => a.filename)).toEqual(['new.html', 'old.html']);
});
it('records external artifacts (image-watcher correlation hook)', () => {
index.recordExternalArtifact(
'agent-1',
'sess-1',
'/p/shots/01-desktop.png',
Date.parse('2026-06-13T05:45:00.000Z')
);
const arts = index.getArtifactsForAgent('agent-1');
expect(arts).toHaveLength(1);
expect(arts[0]).toMatchObject({ source: 'external', kind: 'image', filename: '01-desktop.png' });
});
it('bounds artifacts per agent with FIFO eviction', () => {
for (let i = 0; i < MAX_ARTIFACTS_PER_AGENT + 5; i++) {
emitCall({ tool: 'Write', input: { file_path: `/p/f${i}.html`, content: '' } });
}
expect(index.getArtifactsForAgent('agent-1')).toHaveLength(MAX_ARTIFACTS_PER_AGENT);
// Oldest (f0..f4) evicted; newest retained.
expect(index.getArtifact('agent-1', '/p/f0.html')).toBeUndefined();
expect(index.getArtifact('agent-1', `/p/f${MAX_ARTIFACTS_PER_AGENT + 4}.html`)).toBeDefined();
});
it('stop() detaches from the source; clear() drops artifacts', () => {
emitCall({ tool: 'Write', input: { file_path: '/p/a.html', content: '' } });
index.stop();
emitCall({ tool: 'Write', input: { file_path: '/p/b.html', content: '' } });
expect(index.count()).toBe(1); // second call ignored after stop
index.clear();
expect(index.count()).toBe(0);
});
it('start() is idempotent — no double counting on repeat start with same source', () => {
index.start(source);
index.start(source);
const added: AgentArtifact[] = [];
index.on('artifact:added', (a) => added.push(a));
emitCall({ tool: 'Write', input: { file_path: '/p/a.html', content: '' } });
expect(added).toHaveLength(1);
});
it('clearAgent() drops only the named agent', () => {
emitCall({ agentId: 'a1', tool: 'Write', input: { file_path: '/p/1.html', content: '' } });
emitCall({ agentId: 'a2', tool: 'Write', input: { file_path: '/p/2.html', content: '' } });
index.clearAgent('a1');
expect(index.getArtifactsForAgent('a1')).toHaveLength(0);
expect(index.getArtifactsForAgent('a2')).toHaveLength(1);
});
});