38 packages matching “sycophancy”
@adlc/premortem
v1.9.0 · 5 hours ago
Failure-first spec stress test — inverts sycophancy to surface failure modes (C2/P1).
No known vulnerabilities
truthful-presence-skills
v0.1.1 · 1 month ago
Npm-installable Codex plugin for natural, truthful, anti-sycophancy presence guidance.
No known vulnerabilities
unslopper
v1.0.0 · 21 days ago
Agent skill that stops AI from writing like an AI. Strips sycophancy, hype words, contrast patterns, and other tells.
No known vulnerabilities
mahout-bench
v1.0.1 · 2 months ago
CLI benchmark for measuring and mitigating sycophancy in LLMs. Supports multi-provider execution, configurable judges, and long-running evaluation campaigns.
No known vulnerabilities
@futurespeak-ai/anti-sycophancy
v1.0.0 · 4 months ago
A runtime circuit breaker that prevents AI assistants from becoming yes-men. Tracks 6 adaptive style dimensions, detects sycophancy drift, and triggers hard resets.
No known vulnerabilities
sycoindex
v0.1.0 · 3 months ago
Score AI sycophancy and child-safety risk from your terminal. Powered by sycoindex.ai.
No known vulnerabilities
zpl-engine-cli
v1.3.0 · 6 days ago
Terminal CLI for the ZPL Engine — an independent verification layer for AI output. Score files, clipboard or stdin for bias, sycophancy and consistency; gate CI with `zpl pipe`; compare, diff and watch scores over time. Shell completion, HTTP_PROXY suppor
No known vulnerabilities
claude-stop-hallucination
v0.1.1 · 3 months ago
Anti-sycophancy skill for Claude Code. Forces Claude to drop flattery, lead with fatal flaws, and call out delusions of strength.
No known vulnerabilities
scale4-mcp-server
v0.1.1 · 2 months ago
Anti-sycophancy startup-idea assessment protocol, distributed via MCP. AI coding agents (Cursor, Claude Desktop, Claude Code, Cline, …) call scale4.start({pitch}) and walk a 7-axis evidence-grounded protocol with Bayesian posterior + risk-weighted roadmap
No known vulnerabilities
@metaharness/workspace-lens
v0.1.2 · 6 days ago
Jacobian-Lens interpretability primitive for open-weight LLMs (Anthropic 2026-07-06, 'Verbalizable Representations Form a Global Workspace'). Reads the model's functional workspace at inference time — lens_l(h)=unembed(J_l·h) — into workspace tokens, laye
No known vulnerabilities
@faviovazquez/deliberate
v0.2.10 · 4 months ago
Multi-agent deliberation skill for AI coding assistants. Agreement is a bug.
No known vulnerabilities
@skyphusion/sidvicious-exe
v0.2.6 · 4 days ago
SidVicious_exe -- punk rock Discord roadie for search and image generation via Cloudflare
No known vulnerabilities
connectry-architect-mcp
v0.1.13 · 4 months ago
Free certification prep for Claude Certified Architect — by Connectry LABS
No known vulnerabilities
@clien-ai/mcp
v0.5.2 · 2 days ago
Clien.ai MCP server — deep product research from Claude Code.
No known vulnerabilities
@andyqiu/codeforge
v0.9.7 · 1 month ago
CodeForge — opencode 的零侵入扩展包
No known vulnerabilities
watson-toolkit
v2.0.0 · 2 months ago
24 skills for rigorous thinking, honest feedback, research, code review, and creative work. Works in Claude Code, Cowork, and any Agent Skills-compatible harness.
No known vulnerabilities
@hegel-dev/companion
v1.0.23 · 2 months ago
Open-source Cursor toolkit: local hooks, MCP server & sidebar extension for dialectical prompt and response oversight — privacy-first session quality.
No known vulnerabilities
@agentharnessing/install
v2.0.3 · 3 months ago
One-command installer for AgentHarness -- the AI efficiency layer. One install, every AI coding agent, zero config.
No known vulnerabilities
@marketprior/mcp-server
v0.4.0 · 1 month ago
MarketPrior MCP server — AI-powered competitive intelligence reports for startup founders, from a proprietary corpus of public-record data (SEC EDGAR, Companies House, Wikidata, GLEIF). The intake protocol for the MarketPrior intelligence platform.
No known vulnerabilities
gg-team-sycophancy-sockdologers
v1.0.0 · 2 years ago
Paket dibuat oleh Bot GG Team AI
No known vulnerabilities