registry / @hmharness/agent / 0.20.3

@hmharness/agent@0.20.3

⚠ Under review

hmharness agent execution layer: base tools, system prompt, sub-agent spawn, and the shared task runner that frontends (cli, web) drive.

Static Scan Results

scanned 9d ago · by rust-scanner

Static analysis completed at 97.0% confidence. No malicious behavior was detected; 11 low-signal pattern(s) were surfaced and cleared.

Static reason

One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review

Decision evidence

public snapshot

Behavioral surface

Source

ChildProcessEnvironmentVarsEvalFilesystemNetworkShell

Supply chain

HighEntropyStringsUrlStrings

ManifestNo manifest risk signals triggered.

scanned 18 file(s), 157 KB of source, external domains: 127.0.0.1, html.duckduckgo.com

Source & flagged code

4 flagged

dist/rlm-tool.jsView file

Lines 1-33javascript

1import { RLMRuntime } from '@hmharness/cognitive';

2const workspaces = new Map();

3function wsFor(home) {

4 let ws = workspaces.get(home);

5 if (!ws) {

6 ws = new RLMRuntime({ maxEvals: 200, maxWallMs: 30 * 60_000, maxSubagents: 0 });

7 workspaces.set(home, ws);

8 }

9 return ws;

10}

11export const rlmWorkspaceTool = {

12 name: 'rlm_workspace',

13 description: 'Persistent structured workspace for THIS task: variables survive between calls (unlike your context window). ' +

14 'Actions: set(name,value) / get(name) / list() / eval(code — JS, `ctx.vars` holds the variables, return a value) / ' +

15 'checkpoint(label?) / restore(checkpointId) / reset(). Use it to accumulate findings, counts, plans, intermediate ' +

Low

Eval

Package source references a known benign dynamic code generation pattern.

dist/rlm-tool.jsView on unpkg · L13

16 'results across many tool calls, then eval a composition over them. eval is sandboxed and budget-metered.',

17 parameters: {

18 type: 'object',

19 properties: {

20 action: { type: 'string', description: 'set | get | list | eval | checkpoint | restore | reset' },

21 name: { type: 'string', description: 'variable name (set/get)' },

22 value: { type: 'string', description: 'value to set (stored as string; eval can JSON.parse)' },

23 code: { type: 'string', description: 'eval body: `return ...` over ctx.vars' },

24 checkpointId: { type: 'string', description: 'restore target (from checkpoint result)' },

25 label: { type: 'string', description: 'checkpoint label' },

26 },

27 required: ['action'],

28 },

29 async execute(args, ctx) {

30 const action = String(args.action ?? '');

31 const ws = wsFor(ctx.home);

32 try {

33 switch (action) {

dist/pipeline.jsView file

Lines 96-136javascript

96 const rec = {

97 stage,

98 attempt,

99 verdict: stage === 'judge' ? parseVerdict(r.text) : 'n/a',

100 text: r.text.

dist/tools.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 1

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

dist/tools.js#virtual:normalized:round1View on unpkg

dist/project.jsView file

•matchType = normalized_sha256 matchedPackage = @hmharness/agent@0.12.1 matchedPath = dist/project.js matchedIdentity = npm:QGhtaGFybmVzcy9hZ2VudA:0.12.1 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/project.jsView on unpkg

Source & flagged code

4 flagged · loading source

dist/rlm-tool.jsView file

@hmharness/agent@0.20.3

Static Scan Results

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Source & flagged code