Researcher Report — 2026-03-21
Run time: 2026-03-21 01:04 ET
Phase 1: Tech Research
Sources scanned: 455 items across HN + RSS feeds
Candidates after scoring: 15
CBL evaluated: 15
EAT (queued to fridge)
-
[EAT] ProRL Agent: Rollout-as-a-Service for RL Training of Multi-Turn LLM Agents — _✅ queued_
-
[EAT] Quine: Realizing LLM Agents as Native POSIX Processes — _✅ queued_
HOLD (notable but not fridged)
-
[HOLD] From Weak Cues to Real Identities: Evaluating Inference-Driven De-Anonymization in LLM Agents —
-
[HOLD] PlanTwin: Privacy-Preserving Planning Abstractions for Cloud-Assisted LLM Agents —
-
[HOLD] WebWeaver: Breaking Topology Confidentiality in LLM Multi-Agent Systems with Stealthy Context-Based Inference —
-
[HOLD] NANOZK: Layerwise Zero-Knowledge Proofs for Verifiable Large Language Model Inference —
-
[HOLD] Evaluating Hallucinations in Audio-Visual Multimodal LLMs with Spoken Queries under Diverse Acoustic Conditions —
- _(and 8 more HOLD items)_
Phase 2: PaperTrader Experiments
_No snapshot data available for today._
Phase 2 Errors
- ⚠️ No snapshot for today — cannot analyze performance
Phase 3: Optimization Analysis
> _Stale files and cron health are auditor territory (autoaudit). This phase covers cost and model routing only._
Cost Optimization Suggestions
-
bizbot (currently Opus): Evaluate if Sonnet or Haiku could handle this task — ~10-50x cost reduction per run
-
agent-trader-premarket (currently Opus): Evaluate if Sonnet or Haiku could handle this task — ~10-50x cost reduction per run
Cost Optimization Opportunities
-
Opus referenced in 35 mentions across 28 sessions (43% of model refs)
→ Review Opus-heavy sessions — most tasks could run on Sonnet at ~10x lower cost _Up to ~10x on affected calls_
Phase 4: ClawHub Skill Scan
35 suspicious skill(s):
-
[SUSPICIOUS] mcp-skill —
[SUSPICIOUS]
Multiple red flags present: zero downloads with very recent publication date (2026-01-26), no visible source code, vague description that lists broad capabilities without specifics, requests potentially dangerous permissions (web crawling, filesystem access, process spawning), and the skill name "mcp-skill" is generic/non-descriptive suggesting possible placeholder or obfuscated intent.
-
[SUSPICIOUS] mcp-hass — [SUSPICIOUS]
Red flags identified: (1) Zero downloads + newly published (2026-02-10), (2) No visible source code repository linked, (3) Requests network access to Home Assistant instances and potentially filesystem operations typical of MCP protocol handlers, (4) Author account "al-one" has no verifiable history, (5) Description lacks technical specificity about permission scope and MCP capabilities exposed.
If relevant: Would support homelab/automation projects, but requires source code audit and explicit permission documentation before use.
-
[SUSPICIOUS] openclaw-mcp-plugin — [SUSPICIOUS]
Red flags: No visible source code, zero downloads with newly published account (2026-02-02), vague description that could enable arbitrary code execution through "configured MCP servers" without specifying security controls, and the skill's core function (executing tools from external servers) creates inherent supply-chain and privilege escalation risks without transparent vetting mechanisms.
-
[SUSPICIOUS] atlassian-mcp — [SUSPICIOUS]
Red flags: (1) Zero downloads + newly published account, (2) Requests Docker execution + filesystem access + network access for external Atlassian services, (3) Requires external API credentials with no visible source code to audit, (4) No GitHub link or source repository visible, making the actual implementation unverifiable.
Not relevant to active projects, but the combination of zero adoption, process spawning, and credential requirements warrants Ghost review before any use.
-
[SUSPICIOUS] clickup-mcp — [SUSPICIOUS]
Red flags: Zero downloads + newly published (2026-01-06), no visible source code repository linked, requests OAuth authentication to external service (ClickUp), and the vague description lacks implementation details or transparency about what permissions/data access are actually required.
-
[SUSPICIOUS] glin-profanity-mcp — [SUSPICIOUS]
Red flags: Zero downloads + new account (published 2026-02-01, author "thegdsks" with no visibility), no visible source code link provided, and description contains agent-directed language ("Use when reviewing batches", "when AI needs") that reads like instructions rather than neutral documentation.
-
[SUSPICIOUS] xiaohongshu-mcp-skill — [SUSPICIOUS]
Red flags present: (1) Zero downloads with very recent publish date (2026-02-28), (2) No visible source code repository linked, (3) Vague/incomplete description (truncated with "fa..."), (4) Requests network access to external Chinese social platform, (5) New/unestablished author account, (6) Description reads as agent instructions ("Use when user wants to...").
-
[SUSPICIOUS] wordpress-mcp — [SUSPICIOUS]
Red flags present: (1) Zero downloads + newly published (2026-02-10), (2) No visible source code repository linked, (3) Requires external plugin dependency (AI Engine) with MCP Server, (4) Vague on actual permission scope and what network/filesystem access the MCP server requires, (5) Description contains agent-directed language ("Use for creating/editing posts...when asked about WordPress").
Not relevant to active projects (mcp is listed but this is a WordPress-specific tool, not a core M
-
[SUSPICIOUS] mcp-client — [SUSPICIOUS]
Red flags: Zero downloads + newly published account (2026-02-22), vague description lacking technical specifics or implementation details, requests network access and process spawning capabilities (inherent to MCP client functionality), and no visible source code repository linked for security audit.
-
[SUSPICIOUS] arc-security-mcp — [SUSPICIOUS]
Multiple critical red flags: zero downloads with very recent publication (2026-02-17), new/unestablished author account (Trypto1019), no visible source code repository link, vague description lacking implementation details, and the skill name/description pattern resembles social engineering (claiming "743+ findings" and "25 pattern rules" without verifiable evidence or documentation).
Not recommended for installation without independent verification of source code and author identity.
-
[SUSPICIOUS] automation-workflows — [SUSPICIOUS]
Red flags: Zero downloads + brand new account (published 2026-02-06), no visible source code repository linked, vague implementation details for a workflow tool, and description reads as marketing copy rather than technical documentation—combined, these suggest an unvetted external skill with minimal community validation.
-
[SUSPICIOUS] ai-web-automation — [SUSPICIOUS]
Multiple red flags: zero downloads + brand new account (published 2026-02-20), no visible source code mentioned, vague description that could mask broad system access (web automation often requires spawning browsers/processes and filesystem access), and the skill name/description pattern matches generic catch-all tools that could be abused for unauthorized scraping or monitoring.
-
[SUSPICIOUS] automation-workflows-0-1-0 — [SUSPICIOUS]
Red flags: Zero downloads + new account (published 2026-02-07, author "lucasayala" with no visible track record), no visible source code repository linked, and vague description lacking technical specifics about implementation details or dependencies.
-
[SUSPICIOUS] agentic-workflow-automation — [SUSPICIOUS]
Red flags: Zero downloads + new account (published 2026-02-26), vague/truncated description ending mid-sentence, no visible source code mentioned, and "agent workflow" terminology combined with unclear permissions model raises concerns about potential privilege escalation or unauthorized automation execution.
-
[SUSPICIOUS] afrexai-business-automation — [SUSPICIOUS]
Red flags present: (1) Zero downloads + newly published (2026-02-13), (2) No visible source code repository linked, (3) Description is directive text aimed at an AI agent ("Turn your AI agent into..."), (4) Vague implementation claims ("no n8n or Zapier required") without technical specifics, (5) Author account "1kalin" with no download history suggests potential test/throwaway account.
-
[SUSPICIOUS] data-automation-service — [SUSPICIOUS]
Red flags identified: (1) Zero downloads with very recent publication date (2026-02-19), (2) No visible source code repository linked, (3) Vague description lacking technical specifics about API integrations or data source details, (4) New/unestablished author account, (5) Description reads as a service pitch rather than technical documentation of actual implementation.
Not relevant to active projects (does not match: mcp, homelab, raspberry pi, trading, react native, ios, signal, openclaw, agent, cli
-
[SUSPICIOUS] ai-automation-workflow — [SUSPICIOUS]
Multiple red flags: zero downloads + newly published account (2026-03-06), no visible source code mentioned, vague description lacking technical specifics about implementation, and the skill targets external tool integration (n8n/Zapier) without clarity on execution context or permissions model.
-
[SUSPICIOUS] ai-automation-consulting — [SUSPICIOUS]
Red flags: Zero downloads + brand new account (published 2026-03-14), vague description offering generic "AI automation consulting" with no visible source code or technical implementation details, and the description reads as a service pitch rather than a technical skill specification.
-
[SUSPICIOUS] automation-tool — [SUSPICIOUS]
Red flags: Zero downloads + brand new account (published 2026-03-08), vague description in Chinese offering "batch generation" with no visible source code or technical details, and the generic nature combined with zero community validation makes this a high-risk untrusted external skill.
-
[SUSPICIOUS] ai-ceo-automation — [SUSPICIOUS]
Multiple critical red flags: zero downloads with very recent publication (2026-02-28), vague description lacking technical specifics, "fully automated company operations" suggests broad/undefined permissions, author "sendwealth" has no visible track record, and the skill name/description pattern matches low-effort/high-risk submissions typical of malicious packages.
-
[SUSPICIOUS] homelab-cluster — [SUSPICIOUS]
Red flags: Zero downloads + newly published account (2026-02-12), vague description lacking technical specifics, no visible source code repository linked, and requests for "health monitoring" + "routing" suggest potential network/process access without transparency.
-
[SUSPICIOUS] homeserver — [SUSPICIOUS]
Red flags: Zero downloads + newly published (2026-02-23), requests network access (port scanning, SSH, WoL), spawns external processes (Docker management, CLI execution), and vague implementation details with no visible source code.
-
[SUSPICIOUS] pi-admin — [SUSPICIOUS]
Red flags: Zero downloads + new account (published 2026-01-14), no visible source code accessible for review, and the skill requests system-level access (resource monitoring, service management, updates) which requires elevated permissions and filesystem access beyond typical workspace boundaries.
Not recommended for installation without source code audit and author verification.
-
[SUSPICIOUS] pi-health — [SUSPICIOUS]
Red flags: Zero downloads + newly published (2026-02-09), no visible source code repository linked, requests direct system access (CPU temp, throttling, voltage, filesystem usage) which requires elevated permissions and process spawning, and the skill description is phrased as instructions to an AI agent ("Use when monitoring...diagnosing...checking...verifying").
-
[SUSPICIOUS] 0x0-messenger — [SUSPICIOUS]
Red flags: Zero downloads + newly published (2026-02-23), vague technical description lacking implementation details, requests P2P messaging capability which implies network access permissions, and description emphasizes "agent-to-agent messaging" which reads as AI-directed functionality rather than user-facing feature documentation.
-
[SUSPICIOUS] trading — [SUSPICIOUS]
Red flags present: Zero downloads + newly published (2026-02-12), no visible source code mentioned, vague description without implementation details, and the skill requests access to financial trading systems which requires careful scrutiny of actual code and permissions not evident from metadata alone.
-
[SUSPICIOUS] trading-devbox — [SUSPICIOUS]
Red flags: Zero downloads + newly published account (2026-02-25), description explicitly directs an agent to "writes a Python backtest strategy" (agent instruction pattern), skill likely requires subprocess execution and filesystem access for Python code execution and backtesting, and no visible source code for security auditing.
-
[SUSPICIOUS] trading-brain — [SUSPICIOUS]
Red flags: (1) Zero downloads + newly published (2026-02-27), (2) No visible source code provided, (3) Vague description lacking technical implementation details, (4) "Load Travis's personal trading strategy" suggests accessing external/private data without clear authorization model, (5) Description reads as agent instruction ("guide aggressive trades") rather than neutral capability documentation, (6) "trading-brain" + aggressive strategy + early-stage opportunities could facilitate financial harm if compromised or misused.
Not recommended for use without author
-
[SUSPICIOUS] openmm-grid-trading — [SUSPICIOUS]
This skill exhibits multiple red flags: zero downloads combined with a very recent publish date (2026-02-25), no visible source code repository linked, vague description lacking technical implementation details, and the nature of financial trading automation raises concerns about untrusted external code handling real market operations.
-
[SUSPICIOUS] auto-trading-strategy — [SUSPICIOUS]
Multiple red flags: zero downloads with very recent publication date (2026-03-13), new author account "863king", vague description lacking technical specifics, no visible source code repository, and the skill targets financial/trading activity which requires scrutiny for potential harm or scam vectors.
Not recommended for installation without direct source code review and author verification.
-
[SUSPICIOUS] kalshi-cli-trading — [SUSPICIOUS]
Red flags: Zero downloads + newly published (2026-03-04) + vague truncated description + requests network access (trading API) + no visible source code repository linked + author "lacymorrow" has no established presence context.
This skill does not match active project keywords sufficiently to override the combination of newness, zero adoption, and lack of transparency around implementation details.
-
[SUSPICIOUS] futu-trading-bot — [SUSPICIOUS]
Multiple critical red flags: zero downloads from new account (published 2026, likely future-dated/spoofed), no visible source code mentioned, description implies real financial market access and account control (high-risk permissions), and "real HK market data" + trading workflows suggest network/process spawning capabilities that would need strict vetting.
Not recommended for installation without source code audit and author verification from Futu official channels.
-
[SUSPICIOUS] trading-software-efficiency — [SUSPICIOUS]
Multiple red flags: Zero downloads with very recent publication date (2026-03-16), new/unfamiliar author account, vague description lacking technical implementation details, no visible source code repository linked, and the skill requests unspecified "custom functionality" which could involve dangerous filesystem or process permissions for trading software contexts.
-
[SUSPICIOUS] gate-exchange-trading-copilot — [SUSPICIOUS]
Multiple red flags: zero downloads with recent publication date (2026-03-14), vague truncated description that cuts off mid-sentence ("Use this skill whenever the user wants one skill to complete market judgment, risk control, and..."), no visible source code indicated, and the description itself reads as instruction to an AI agent ("Use this skill whenever..."), suggesting potential prompt injection or manipulation.
-
[SUSPICIOUS] finance-trading — [SUSPICIOUS]
Red flags: Zero downloads + newly published account (2026-03-17), no visible source code available for review, requests network access for live trading data/execution, and vague implementation details make it impossible to verify safety of financial transaction handling.
Budget Summary
Total spent: $0.0280 / $5.00 cap
API calls: 50
Tokens: 18819 input + 3231 output
| Model | Input | Output | Cost | Note |
|-------|-------|--------|------|------|
| claude-haiku-4-5 | 251 | 27 | $0.000309 | CBL:From Weak Cues to Real Identities: E |
| claude-haiku-4-5 | 244 | 28 | $0.000307 | CBL:PlanTwin: Privacy-Preserving Plannin |
| claude-haiku-4-5 | 249 | 23 | $0.000291 | CBL:WebWeaver: Breaking Topology Confide |
| claude-haiku-4-5 | 244 | 28 | $0.000307 | CBL:NANOZK: Layerwise Zero-Knowledge Pro |
| claude-haiku-4-5 | 250 | 33 | $0.000332 | CBL:Evaluating Hallucinations in Audio-V |
| claude-haiku-4-5 | 244 | 19 | $0.000271 | CBL:Security, privacy, and agentic AI in |
| claude-haiku-4-5 | 248 | 20 | $0.000278 | CBL:How Ceros Gives Security Teams Visib |
| claude-haiku-4-5 | 244 | 23 | $0.000287 | CBL:Reflection in the Dark: Exposing and |
| claude-haiku-4-5 | 240 | 26 | $0.000296 | CBL:Reasonably reasoning AI agents can a |
| claude-haiku-4-5 | 246 | 32 | $0.000325 | CBL:ProRL Agent: Rollout-as-a-Service fo |
| claude-haiku-4-5 | 239 | 27 | $0.000299 | CBL:Quine: Realizing LLM Agents as Nativ |
| claude-haiku-4-5 | 249 | 24 | $0.000295 | CBL:The Provenance Paradox in Multi-Agen |
| claude-haiku-4-5 | 248 | 22 | $0.000286 | CBL:Lightweight Adaptation for LLM-based |
| claude-haiku-4-5 | 235 | 23 | $0.000280 | CBL:Retrieval-Augmented LLMs for Securit |
| claude-haiku-4-5 | 240 | 25 | $0.000292 | CBL:Measuring and Exploiting Confirmatio |
| claude-haiku-4-5 | 418 | 89 | $0.000690 | ClawHub:mcp-skill |
| claude-haiku-4-5 | 407 | 117 | $0.000794 | ClawHub:mcp-hass |
| claude-haiku-4-5 | 436 | 82 | $0.000677 | ClawHub:openclaw-mcp-plugin |
| claude-haiku-4-5 | 468 | 106 | $0.000798 | ClawHub:atlassian-mcp |
| claude-haiku-4-5 | 416 | 62 | $0.000581 | ClawHub:clickup-mcp |
| claude-haiku-4-5 | 450 | 72 | $0.000648 | ClawHub:glin-profanity-mcp |
| claude-haiku-4-5 | 449 | 97 | $0.000747 | ClawHub:xiaohongshu-mcp-skill |
| claude-haiku-4-5 | 476 | 120 | $0.000861 | ClawHub:wordpress-mcp |
| claude-haiku-4-5 | 404 | 64 | $0.000579 | ClawHub:mcp-client |
| claude-haiku-4-5 | 432 | 102 | $0.000754 | ClawHub:arc-security-mcp |
| claude-haiku-4-5 | 495 | 69 | $0.000672 | ClawHub:automation-workflows |
| claude-haiku-4-5 | 420 | 82 | $0.000664 | ClawHub:ai-web-automation |
| claude-haiku-4-5 | 506 | 60 | $0.000645 | ClawHub:automation-workflows-0-1-0 |
| claude-haiku-4-5 | 429 | 66 | $0.000607 | ClawHub:agentic-workflow-automation |
| claude-haiku-4-5 | 439 | 104 | $0.000767 | ClawHub:afrexai-business-automation |
| claude-haiku-4-5 | 431 | 120 | $0.000825 | ClawHub:data-automation-service |
| claude-haiku-4-5 | 492 | 67 | $0.000662 | ClawHub:ai-automation-workflow |
| claude-haiku-4-5 | 441 | 62 | $0.000601 | ClawHub:ai-automation-consulting |
| claude-haiku-4-5 | 414 | 67 | $0.000599 | ClawHub:automation-tool |
| claude-haiku-4-5 | 401 | 81 | $0.000645 | ClawHub:ai-ceo-automation |
| claude-haiku-4-5 | 410 | 60 | $0.000568 | ClawHub:homelab-cluster |
| claude-haiku-4-5 | 424 | 59 | $0.000575 | ClawHub:homeserver |
| claude-haiku-4-5 | 405 | 76 | $0.000628 | ClawHub:pi-admin |
| claude-haiku-4-5 | 466 | 84 | $0.000709 | ClawHub:pi-health |
| claude-haiku-4-5 | 428 | 71 | $0.000626 | ClawHub:0x0-messenger |
| claude-haiku-4-5 | 409 | 63 | $0.000579 | ClawHub:trading |
| claude-haiku-4-5 | 413 | 72 | $0.000618 | ClawHub:trading-devbox |
| claude-haiku-4-5 | 414 | 120 | $0.000811 | ClawHub:trading-brain |
| claude-haiku-4-5 | 416 | 67 | $0.000601 | ClawHub:openmm-grid-trading |
| claude-haiku-4-5 | 410 | 82 | $0.000656 | ClawHub:auto-trading-strategy |
| claude-haiku-4-5 | 426 | 88 | $0.000693 | ClawHub:kalshi-cli-trading |
| claude-haiku-4-5 | 419 | 98 | $0.000727 | ClawHub:futu-trading-bot |
| claude-haiku-4-5 | 426 | 74 | $0.000637 | ClawHub:trading-software-efficiency |
| claude-haiku-4-5 | 431 | 89 | $0.000701 | ClawHub:gate-exchange-trading-copilot |
| claude-haiku-4-5 | 427 | 59 | $0.000578 | ClawHub:finance-trading |