opencodekit 0.17.13 → 0.18.0

package/dist/template/.opencode/tool/observation.ts

@@ -1,357 +0,0 @@
-import { tool } from "@opencode-ai/plugin";
-import {
-	type ConfidenceLevel,
-	type ObservationInput,
-	type ObservationType,
-	storeObservation,
-} from "../plugin/lib/memory-db.js";
-
-const TYPE_ICONS: Record<ObservationType, string> = {
-	decision: "🎯",
-	bugfix: "🐛",
-	feature: "✨",
-	pattern: "🔄",
-	discovery: "💡",
-	learning: "📚",
-	warning: "⚠️",
-};
-
-const CONFIDENCE_ICONS: Record<ConfidenceLevel, string> = {
-	high: "🟢",
-	medium: "🟡",
-	low: "🔴",
-};
-
-// Patterns to detect file references in observation content
-const FILE_PATTERNS = [
-	// file:line format (e.g., src/auth.ts:42)
-	/(?:^|\s)([a-zA-Z0-9_\-./]+\.[a-zA-Z]{2,4}):(\d+)/g,
-	// backtick file paths (e.g., `src/auth.ts`)
-	/`([a-zA-Z0-9_\-./]+\.[a-zA-Z]{2,4})`/g,
-	// common source paths
-	/(?:^|\s)(src\/[a-zA-Z0-9_\-./]+\.[a-zA-Z]{2,4})/g,
-	/(?:^|\s)(\.opencode\/[a-zA-Z0-9_\-./]+\.[a-zA-Z]{2,4})/g,
-];
-
-interface FileReference {
-	file: string;
-	line?: number;
-}
-
-// Extract file references from observation content
-function extractFileReferences(content: string): FileReference[] {
-	const refs: FileReference[] = [];
-	const seen = new Set<string>();
-
-	for (const pattern of FILE_PATTERNS) {
-		// Reset regex state
-		pattern.lastIndex = 0;
-		let match = pattern.exec(content);
-
-		while (match !== null) {
-			const file = match[1];
-			const line = match[2] ? Number.parseInt(match[2], 10) : undefined;
-			const key = `${file}:${line || ""}`;
-
-			if (!seen.has(key) && !file.includes("node_modules")) {
-				seen.add(key);
-				refs.push({ file, line });
-			}
-			match = pattern.exec(content);
-		}
-	}
-
-	return refs;
-}
-
-export default tool({
-	description: `Create a structured observation for future reference.
-	
-	Purpose:
-	- Capture decisions, bugs, features, patterns, discoveries, learnings, or warnings
-	- Auto-detects file references from content (file:line, \`path\`, src/, .opencode/)
-	- Stores in SQLite with FTS5 index for fast search
-	- Supports enhanced schema: facts, subtitle, files_read/files_modified
-	
-	Confidence guidance:
-	- high: verified by tests, logs, or direct inspection (default)
-	- medium: likely, but not fully verified
-	- low: uncertain or speculative
-	
-	Type-specific examples:
-	decision
-	observation({
-	  type: "decision",
-	  title: "Use JWT for auth",
-	  narrative: "Chose JWT for stateless auth across services.",
-	  facts: "stateless, scalable",
-	  concepts: "authentication, jwt",
-	  confidence: "high"
-	})
-	
-	bugfix
-	observation({
-	  type: "bugfix",
-	  title: "Fix null pointer on login",
-	  narrative: "Guarded optional user in src/auth.ts:42 to prevent crash.",
-	  files_modified: "src/auth.ts",
-	  concepts: "auth, null-check",
-	  confidence: "high"
-	})
-	
-	feature
-	observation({
-	  type: "feature",
-	  title: "Add CLI --dry-run",
-	  narrative: "Introduce dry-run mode to show planned changes without writing.",
-	  files_modified: "src/commands/init.ts",
-	  concepts: "cli, ux",
-	  confidence: "medium"
-	})
-	
-	pattern
-	observation({
-	  type: "pattern",
-	  title: "Use zod for input validation",
-	  narrative: "All command inputs validated with zod schemas before execute.",
-	  concepts: "validation, zod",
-	  confidence: "high"
-	})
-	
-	discovery
-	observation({
-	  type: "discovery",
-	  title: "Build copies .opencode/ to dist/template/",
-	  narrative: "Found rsync step in build.ts that bundles .opencode/.",
-	  files_read: "build.ts",
-	  confidence: "high"
-	})
-	
-	learning
-	observation({
-	  type: "learning",
-	  title: "Bun test respects --watch",
-	  narrative: "Observed bun test --watch keeps runner active during edits.",
-	  confidence: "medium"
-	})
-	
-	warning
-	observation({
-	  type: "warning",
-	  title: "Do not edit dist/ directly",
-	  narrative: "dist/ is built output and overwritten on build.",
-	  concepts: "build, generated",
-	  confidence: "high"
-	})`,
-	args: {
-		type: tool.schema
-			.string()
-			.describe(
-				"Observation type: decision, bugfix, feature, pattern, discovery, learning, warning",
-			),
-		title: tool.schema.string().describe("Brief title for the observation"),
-		subtitle: tool.schema
-			.string()
-			.optional()
-			.describe("Optional subtitle or tagline"),
-		facts: tool.schema
-			.string()
-			.optional()
-			.describe("Comma-separated key facts (e.g., 'stateless, scalable')"),
-		narrative: tool.schema
-			.string()
-			.optional()
-			.describe("Detailed observation content with context"),
-		content: tool.schema
-			.string()
-			.optional()
-			.describe(
-				"DEPRECATED: Use 'narrative' instead. Alias for backward compat.",
-			),
-		concepts: tool.schema
-			.string()
-			.optional()
-			.describe(
-				"Comma-separated concept tags (e.g., 'authentication, oauth, security')",
-			),
-		files_read: tool.schema
-			.string()
-			.optional()
-			.describe("Comma-separated files that were read (e.g., 'src/auth.ts')"),
-		files_modified: tool.schema
-			.string()
-			.optional()
-			.describe("Comma-separated files that were modified"),
-		files: tool.schema
-			.string()
-			.optional()
-			.describe(
-				"DEPRECATED: Use 'files_modified' instead. Alias for backward compat.",
-			),
-		bead_id: tool.schema
-			.string()
-			.optional()
-			.describe("Related bead ID for traceability"),
-		confidence: tool.schema
-			.string()
-			.optional()
-			.describe(
-				"Confidence level: high (verified), medium (likely), low (uncertain). Defaults to high.",
-			),
-		supersedes: tool.schema
-			.string()
-			.optional()
-			.describe(
-				"ID or filename of observation this supersedes (for contradiction handling)",
-			),
-	},
-	execute: async (args: {
-		type: string;
-		title: string;
-		subtitle?: string;
-		facts?: string;
-		narrative?: string;
-		content?: string;
-		concepts?: string;
-		files_read?: string;
-		files_modified?: string;
-		files?: string;
-		bead_id?: string;
-		confidence?: string;
-		supersedes?: string;
-	}) => {
-		// Validate type
-		const validTypes: ObservationType[] = [
-			"decision",
-			"bugfix",
-			"feature",
-			"pattern",
-			"discovery",
-			"learning",
-			"warning",
-		];
-		const obsType = args.type.toLowerCase() as ObservationType;
-		if (!validTypes.includes(obsType)) {
-			return `Error: Invalid observation type '${args.type}'.\nValid types: ${validTypes.join(", ")}`;
-		}
-
-		// Validate confidence level
-		const validConfidence: ConfidenceLevel[] = ["high", "medium", "low"];
-		const confidence = (args.confidence?.toLowerCase() ||
-			"high") as ConfidenceLevel;
-		if (!validConfidence.includes(confidence)) {
-			return `Error: Invalid confidence level '${args.confidence}'.\nValid levels: ${validConfidence.join(", ")}`;
-		}
-
-		// Handle deprecated fields (backward compat)
-		const narrative = args.narrative || args.content || "";
-		const filesModifiedRaw = args.files_modified || args.files || "";
-
-		// Parse arrays from comma-separated strings
-		const facts = args.facts
-			? args.facts
-					.split(",")
-					.map((f) => f.trim())
-					.filter(Boolean)
-			: [];
-		const concepts = args.concepts
-			? args.concepts
-					.split(",")
-					.map((c) => c.trim())
-					.filter(Boolean)
-			: [];
-		let filesRead = args.files_read
-			? args.files_read
-					.split(",")
-					.map((f) => f.trim())
-					.filter(Boolean)
-			: [];
-		const filesModified = filesModifiedRaw
-			? filesModifiedRaw
-					.split(",")
-					.map((f) => f.trim())
-					.filter(Boolean)
-			: [];
-
-		// Auto-detect file references from narrative
-		const detectedRefs = extractFileReferences(narrative);
-		const detectedFiles = detectedRefs.map((r) => r.file);
-
-		// Merge detected files with explicitly provided files
-		filesRead = [...new Set([...filesRead, ...detectedFiles])];
-
-		// Parse supersedes (could be numeric ID or filename)
-		let supersedesId: number | undefined;
-		if (args.supersedes) {
-			const parsed = Number.parseInt(args.supersedes, 10);
-			if (!Number.isNaN(parsed)) {
-				supersedesId = parsed;
-			}
-		}
-
-		// Prepare observation input
-		const input: ObservationInput = {
-			type: obsType,
-			title: args.title,
-			subtitle: args.subtitle,
-			facts: facts.length > 0 ? facts : undefined,
-			narrative: narrative || undefined,
-			concepts: concepts.length > 0 ? concepts : undefined,
-			files_read: filesRead.length > 0 ? filesRead : undefined,
-			files_modified: filesModified.length > 0 ? filesModified : undefined,
-			confidence,
-			bead_id: args.bead_id,
-			supersedes: supersedesId,
-		};
-
-		try {
-			// Store in SQLite (single source of truth)
-			const sqliteId = storeObservation(input);
-
-			// Update bead notes if bead_id provided
-			let beadUpdate = "";
-			if (args.bead_id) {
-				try {
-					const { execSync } = await import("node:child_process");
-					const noteContent = `${TYPE_ICONS[obsType]} ${obsType}: ${args.title}`;
-					execSync(
-						`br edit ${args.bead_id} --note "${noteContent.replace(/"/g, '\\"')}"`,
-						{
-							cwd: process.cwd(),
-							encoding: "utf-8",
-							timeout: 5000,
-						},
-					);
-					beadUpdate = `\nBead updated: ${args.bead_id}`;
-				} catch {
-					beadUpdate = `\nWarning: Could not update bead ${args.bead_id}`;
-				}
-			}
-
-			// Build output
-			const icon = TYPE_ICONS[obsType];
-			const confIcon = CONFIDENCE_ICONS[confidence];
-
-			let output = `✓ Observation #${sqliteId} saved\n\n`;
-			output += `**Type**: ${icon} ${obsType}\n`;
-			output += `**Title**: ${args.title}\n`;
-			output += `**Confidence**: ${confIcon} ${confidence}\n`;
-
-			if (concepts.length > 0) {
-				output += `**Concepts**: ${concepts.join(", ")}\n`;
-			}
-			if (facts.length > 0) {
-				output += `**Facts**: ${facts.length} extracted\n`;
-			}
-
-			output += beadUpdate;
-
-			return output;
-		} catch (error) {
-			if (error instanceof Error) {
-				return `Error saving observation: ${error.message}`;
-			}
-			return "Unknown error saving observation";
-		}
-	},
-});