@henols/vice-mcp 0.2.0 → 0.2.1

package/THIRD-PARTY-NOTICES.md

@@ -75,6 +75,29 @@ package**, so ACME's licence does not attach to anything shipped. ACME never
 appears in `.claude/mcp/vice/package.json`'s `files[]`, `dependencies`, or
 `devDependencies` — it is an apt/CI-installed tool, never an npm package.
 
+## Build/CI tools — not incorporated: regenerator2000
+
+regenerator2000 is invoked as an **external CLI subprocess** (Phase 10's
+`R2000-09` bootstrap and `R2000-06` reassembly proof) against a real,
+locally-installed `regenerator2000` binary. **No regenerator2000 source,
+data table, or output is included in this repository or in either
+published package**, so its licence does not attach to anything shipped.
+
+Its licence is **`MIT OR Apache-2.0`** — dual, at the user's option —
+confirmed from the crate's own crates.io licence field, with both
+`LICENSE-MIT` and `LICENSE-APACHE` shipping in the crate. (Do not read this
+as Apache-2.0 alone: that stale, single-licence claim is what `R2000-03`'s
+own requirement text and `.planning/notes/regenerator2000-integration.md`
+still carried before Phase 10 corrected it here.)
+
+Verified version `0.9.20`, published 2026-07-11 by `ricardoquesada`
+(matching the linked GitHub repository owner), checked 2026-08-20 — so this
+provenance claim is re-checkable against a specific release.
+
+regenerator2000 never appears in `.claude/mcp/vice/package.json`'s
+`files[]`, `dependencies`, or `devDependencies` — it is a `cargo
+install`-provided tool, never an npm package.
+
 ## Explicitly NOT a source: VICE
 
 VICE is GPL-2 and this repository is MIT. **No opcode fact, protocol

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@henols/vice-mcp",
-  "version": "0.2.0",
+  "version": "0.2.1",
   "description": "VICE emulator MCP server for C64 reverse-engineering: a stdio MCP server that proxies vice tools to a host VICE MCP server.",
   "type": "module",
   "bin": {
@@ -53,6 +53,20 @@
     "stock-run-until.ts",
     "stock-diagnose.ts",
     "stock-recycle.ts",
+    "r2000-launch.ts",
+    "r2000-project.ts",
+    "r2000-d64.ts",
+    "r2000-cli.ts",
+    "r2000-verify.ts",
+    "r2000-mcp-client.ts",
+    "r2000-tools.ts",
+    "r2000-symbols.ts",
+    "r2000-regbits-gen.ts",
+    "r2000-regbits.json",
+    "r2000-enum-gen.ts",
+    "r2000-acme-ident.ts",
+    "r2000-confidence.ts",
+    "r2000-memmap-render.ts",
     "resources",
     "tools-manifest.json",
     "tools-manifest.stock.json",

package/r2000-acme-ident.ts

@@ -0,0 +1,97 @@
+#!/usr/bin/env node
+// r2000-acme-ident.ts -- the ONE authoritative place in this repo for the
+// ACME identifier policy (T-11-ENUM-NAME / T-11-NAME-INJECT): what makes a
+// string a legal ACME symbol/label name, and the single check function that
+// decides it.
+//
+// WHY THIS MODULE EXISTS: this policy started life inside r2000-enum-gen.ts,
+// consumed only by its own `createOrUpdateEnum()`/`sanitizeVariantMap()`.
+// T-11-NAME-INJECT widened the finding to a SECOND entry route --
+// `r2000_set_label_name` (both outer and batch-inner) in r2000-tools.ts, and
+// `importLabels()` in r2000-symbols.ts -- and `r2000-enum-gen.ts` statically
+// imports `runR2000Tool` FROM `r2000-tools.ts`, so `r2000-tools.ts` cannot
+// import the policy back from `r2000-enum-gen.ts` without forming a module
+// cycle. This module has no import from anywhere else in this repo, so
+// every one of those consumers (and any future one) can import it directly.
+//
+// WHAT THIS IS THE ONE AUTHORITATIVE PLACE FOR: `MAX_ACME_IDENTIFIER_LENGTH`,
+// `ACME_IDENT_RE`, `ACME_RESERVED_MNEMONICS` and `assertLegalAcmeIdentifier()`
+// -- the complete, only definition of "legal ACME identifier" in this repo.
+// `r2000-enum-gen.ts` re-exports `assertLegalAcmeIdentifier` /
+// `MAX_ACME_IDENTIFIER_LENGTH` from here so its own existing consumers and
+// tests keep their current import path; it does not hold a second copy.
+//
+// WHAT NOT TO DO, named concretely:
+//   - Never add a second identifier regex anywhere in this repo. The
+//     `r2000-regbits-gen.ts` / `r2000-regbits.test.ts` copies of
+//     `ACME_IDENT_RE` predate this module and are out of this plan's scope
+//     (260821-a86) to consolidate -- but no NEW copy should be added; import
+//     this module's `assertLegalAcmeIdentifier()` instead.
+//   - Never sanitize, quote, or auto-correct an illegal identifier here or
+//     in any caller. This function's contract is REJECT, per
+//     T-11-REGBITS-PROSE and T-11-ENUM-NAME precedent: a malformed name is a
+//     bug to surface, never silently rewritten into something else -- the
+//     caller-visible name must never diverge from what actually gets
+//     exported into ACME source.
+//   - Never import anything from this repo into this module. It must stay
+//     importable by both `r2000-tools.ts` and `r2000-enum-gen.ts` (which
+//     imports `runR2000Tool` FROM `r2000-tools.ts`) without a cycle.
+export const MAX_ACME_IDENTIFIER_LENGTH = 200;
+
+const ACME_IDENT_RE = /^[A-Za-z_][A-Za-z0-9_]*$/;
+
+/**
+ * The 6502/6510 instruction mnemonics -- the ONLY category of reserved word
+ * actually measured to collide with a bare identifier in real ACME 0.97.
+ * Verified live on this host: `LDA = $05` is REJECTED ("No value given." /
+ * "Garbage data at end of statement" -- ACME parses `LDA` positionally as
+ * the opcode, not as an assignable symbol), while `A = $05` is ACCEPTED
+ * (register-letter shorthand is not reserved as a bare symbol name). This
+ * list is therefore a measured, not a guessed, reservation.
+ */
+const ACME_RESERVED_MNEMONICS: ReadonlySet<string> = new Set([
+  "ADC", "AND", "ASL", "BCC", "BCS", "BEQ", "BIT", "BMI", "BNE", "BPL", "BRK", "BVC",
+  "BVS", "CLC", "CLD", "CLI", "CLV", "CMP", "CPX", "CPY", "DEC", "DEX", "DEY", "EOR",
+  "INC", "INX", "INY", "JMP", "JSR", "LDA", "LDX", "LDY", "LSR", "NOP", "ORA", "PHA",
+  "PHP", "PLA", "PLP", "ROL", "ROR", "RTI", "RTS", "SBC", "SEC", "SED", "SEI", "STA",
+  "STX", "STY", "TAX", "TAY", "TSX", "TXA", "TXS", "TYA",
+]);
+
+/**
+ * Refuses `id` (naming it as `what` in the thrown message) unless it is a
+ * legal ACME identifier: matches `^[A-Za-z_][A-Za-z0-9_]*$`, is no longer
+ * than `MAX_ACME_IDENTIFIER_LENGTH`, and does not collide (case-
+ * insensitively) with a reserved 6502/6510 mnemonic (see
+ * `ACME_RESERVED_MNEMONICS`'s own header comment for how that specific list
+ * was measured, not assumed).
+ *
+ * T-11-ENUM-NAME (the highest-value threat in this phase): regenerator2000
+ * validates only the ENUM name server-side
+ * (`app_state.rs:443`, `validate_new_enum_name`) and performs ZERO
+ * validation on variant names -- they flow straight into
+ * `format!("{}_{}", enum_name, variant)` at export time
+ * (`formatter_acme.rs:367-369`). This function is called on BOTH the enum
+ * name and every variant name, and is called BEFORE any
+ * `r2000_create_project_enum`/`r2000_update_project_enum` call reaches
+ * `runR2000Tool()` -- proven zero-spawn in `r2000-enum-gen.test.ts`.
+ */
+export function assertLegalAcmeIdentifier(id: string, what: string): void {
+  if (id.length === 0) {
+    throw new Error(`${what}: identifier must not be empty`);
+  }
+  if (id.length > MAX_ACME_IDENTIFIER_LENGTH) {
+    throw new Error(
+      `${what}: identifier "${id.slice(0, 40)}..." is ${id.length} characters, exceeding the ` +
+        `${MAX_ACME_IDENTIFIER_LENGTH}-character ceiling`,
+    );
+  }
+  if (!ACME_IDENT_RE.test(id)) {
+    throw new Error(`${what}: "${JSON.stringify(id)}" is not a legal ACME identifier -- must match ${ACME_IDENT_RE}`);
+  }
+  if (ACME_RESERVED_MNEMONICS.has(id.toUpperCase())) {
+    throw new Error(
+      `${what}: "${id}" collides with the reserved 6502/6510 mnemonic ${id.toUpperCase()} (verified rejected ` +
+        `by real ACME 0.97 -- see this module's ACME_RESERVED_MNEMONICS header comment)`,
+    );
+  }
+}