LLM Mart Basic

Propagates dimensional annotations through arithmetic and call chains, reporting mismatches found during propagation

Validates dimensional consistency and detects dimensional bugs in annotated code

Analyzes data flow from source to vulnerability sink, mapping trust boundaries, API contracts, environment protections, and cross-references. Spawned by fp-check during Phase 1 verification.

Verifies whether a suspected vulnerability is actually exploitable by proving attacker control, mathematical bounds, and race condition feasibility. Spawned by fp-check during Phase 2 verification.

Creates proof-of-concept exploits (pseudocode, executable, and unit tests) demonstrating a verified vulnerability, plus negative PoCs showing exploit preconditions. Spawned by fp-check during Phase 4 verification.

Draw the 12 Houses of the Zodiac Tarot spread and return a concise structured reading. Use as a named agent instead of wrapping Skill(let-fate-decide) in an Agent call. Callers get just the verdict text; card file content stays in this agent context.

Deduplication judge for the rust-review pipeline. Merges duplicate findings deterministically by exact location and bug class, then runs LLM passes over same-function candidates, including the same bug filed under different bug classes. Spawned by the rust-review skill orchestrat

Second-stage judge in the rust-review pipeline. Runs after dedup-judge on merged primaries only. Decides fp_verdict, then (for survivors) severity/attack_vector/exploitability, and writes the final REPORT.md + REPORT.sarif. Spawned by the rust-review skill orchestrator only.

Runs one assigned rust-review cluster task and writes finding files to the run's output directory. Spawned by the rust-review skill orchestrator only.

Evaluates APIs, configurations, and library interfaces for misuse resistance and footgun potential. Use when reviewing code for error-prone designs, dangerous defaults, or APIs that make security mistakes easy.

Checks one documented requirement against the code that should implement it, and returns a verdict with the lines that evidence it. Writes its analysis to disk and returns a compact record. Use for a single requirement; use the spec-compliance workflow for a whole document.

Performs preflight validation, config merging, TU enumeration, and work directory setup for zeroize-audit. Produces merged-config.yaml, preflight.json, and orchestrator-state.json.

Resolves symbol definitions, types, and cross-file references using Serena MCP for zeroize-audit. Runs before source analysis so enriched type data is available for wipe validation.

Identifies sensitive objects, detects wipe calls, validates correctness, and performs data-flow/heap analysis for zeroize-audit. Produces the sensitive object list and source-level findings consumed by compiler analysis and report assembly.

Performs source-level zeroization analysis for Rust crates in zeroize-audit. Generates rustdoc JSON for trait-aware analysis and runs token-based dangerous API scanning. Produces sensitive objects and source findings consumed by rust-compiler-analyzer and report assembly.

Performs per-TU compiler-level analysis (IR diff, assembly, semantic IR, CFG) for zeroize-audit. One instance runs per translation unit, enabling parallel execution across TUs.

Performs crate-level MIR and LLVM IR analysis for Rust in zeroize-audit. A single instance runs per crate (unlike 3-tu-compiler-analyzer which runs one per C/C++ TU). Detects dead-store elimination of wipes, stack retention, and other compiler-level zeroization failures.

Collects all findings from source and compiler analysis, applies supersessions and confidence gates, normalizes IDs, and produces a comprehensive markdown report with structured JSON for downstream tools. Supports dual-mode invocation: interim (findings.json only) and final (merg

Crafts bespoke proof-of-concept programs demonstrating that zeroize-audit findings are exploitable. Reads source code and finding details to generate tailored PoCs — each PoC is individually written, not templated. Each PoC exits 0 if the secret persists or 1 if wiped. Mandatory

Compiles and runs all PoCs for zeroize-audit findings. Produces poc_validation_results.json consumed by the verification agent and the orchestrator.

Fourteen posts of being wrong in production, compressed to checkboxes

Healthy nodes, a quiet network, 300 restarts in three days, and a latency budget measured in milliseconds

Discovery worked. Ping worked. Every TCP connection timed out, and later the tunnel only worked when someone had a terminal open.

Every VM came back. The cluster did not. Declarative systems converge on config, and the datapath isn't config.

A surprising share of AI-in-the-terminal failures aren't the AI. They're zsh, and a version of bash from 2006.

A Claude Code plugin turns standalone project configuration into a namespaced, installable extension that teams and communities can update as one unit.

None of the safety came from the model. It came from six boring habits.

Skills package instructions and references. Subagents run work in a separate context and return results. They solve different problems and can be composed deliberately.

Six hours in, one step left, everything green, and the incident that didn't happen

CLAUDE.md carries persistent project context. Skills load reusable procedures when relevant. Separating stable facts from task-specific workflows keeps both easier to maintain.

Twenty minutes recovering secrets that never existed, and the one sentence from a human that ended it

An API request routing a model's tool call through an approval gate to a remote MCP server

31 config keys, two audits, and why the first one was wrong in both directions

The official MCP Registry stores standardized server metadata rather than package code. Publishers verify a namespace, describe installation or remote access, and submit immutable versions.

Everyone looks at the Dockerfile. The file that actually leaked the key was the project file.

Remote MCP authorization uses established OAuth standards, but secure integration still requires issuer validation, least-privilege scopes, protected token handling, and server-side enforcement.

"Copy it over and switch the reference" is two steps, and the outage lives in the one nobody checks

stdio fits local processes and prototypes. Streamable HTTP fits hosted services and shared integrations. The right choice follows where the capability runs and who must reach it.

The most important rule wasn't about what I could change. It was about what I was allowed to display.

Tools perform operations, resources expose readable context, and prompts provide reusable templates. Choosing the correct primitive makes an MCP server easier to understand and govern.

"Tiered plugin quality review: branch (quick gates),

Check GitHub Discussions for highly voted learnings and promote them to Issues.

Evaluate Claude Code rules in .claude/rules/ directories for quality

Audit skill quality, frontmatter compliance, token efficiency, and activation reliability. Recommends improvements.

Test Claude Code skills using RED/GREEN/REFACTOR TDD phases in fresh subagents to prevent priming bias.

Validate hooks for security, performance, and SDK compliance

Validate plugin structure, schema, and naming. Use for plugin creation, debugging, or verification.

Initialize projects with architecture-aware templates using paradigm research and selection guidance for the target domain.

Generate an implementation plan with system architecture design and dependency-ordered task breakdown from a specification.

Guide project ideation through Socratic questioning to generate briefs with validated approaches and decision rationale.

Execute implementation plan systematically with progress tracking and checkpoint validation

Run full attune lifecycle as a mission with state detection and phase routing

Initialize a new project with git setup, CI/CD workflows, pre-commit hooks, Makefiles, and language-specific tooling.

Build a project skill library under .claude/skills/ as a resumable mission: discover, author in parallel, review adversarially.

Create specs from project briefs with acceptance criteria and testable requirements

Update existing project configurations to current best practices with selective component upgrades

Validate project structure and configurations against best practices with detailed issue reporting

Convene a multi-LLM expert panel to pressure-test strategic decisions with adversarial review and reversibility assessment.

Generate visual diagrams of codebase structure using Mermaid Chart MCP rendering.

Audit codebase for AI-generated code quality issues (vibe coding, Tab bloat, slop)

Okou connects to the tools your team already uses and does the work — across marketing, sales, engineering, and operations, under your control.

An open-source Digital Worker platform for reliable execution, continuous co-evolution, and building Enterprise AI assets.

A curated list of AI Agent evolution, memory systems, multi-agent architectures, and self-improvement projects. | evomap.ai

A governance harness for AI coding.

DeepSeek Harness 手机版：可直接安装的 Android APK，AI 免 Root 操作手机（Shizuku/root 可选），文件编辑只需所有文件访问权限，前台保活 + AI 通知

Local-first, governed AI agent runtime for Python — embed it in your app, or run it as a CLI or ACP server. Permissions, MCP, memory and audit replay built in.

Orbi — the factory that builds and operates AI software factories. GitHub Issues in, releases and runnable system out

Open-source AI agent harness in native Rust — GUI, CLI, headless, and webapp from one binary. Multi-provider, MCP, skills, plugins, agent teams.

The SDK for browser agents. Interact, search, extract, and fetch any site reliably across the web

Agent-native shopping for extreme value: verifiable same-product price evidence, checkout, orders, delivery, and after-sales.

The extensible power-user platform for Google Antigravity. Adds a native In-App Browser, animated Desktop Pets, revamped Gemini UI, and custom BYOK Gemini Pro k…

Open-source, single-binary, self-hosted AI agent — your models and data stay on your machine. A coding agent on par with Claude Code and a personal assistant li…

Your pocket agent. Local-first AI agents on iOS and Android — real workspaces, tool execution with approvals, and your choice of model (DSH · Claude Code · Code…

Bridge between QQ (SnowLuma OneBot v11) and DeepSeek Harness agents: social simulation, safe MCP tools, slang learning and more.

A 7×6 framework for agent architecture. 28 patterns, each placed at a coordinate, runnable Python code with verified engineering slices from Claude Code, Aider,…

针对 AI 自动化渗透 Agent 的新一代反制蜜罐，通过反向代理将API密饵载入真实业务、反向提示词注入等方式反制自动化渗透 Agent，实现多款主流通用Agent的反制上线控制。

Open-source, local-first AI agent for coding and real work. BYOK models, MCP, skills, plugins, workflows, and private knowledge bases.

A free AI-agent toolbox for Android, 一站式安卓AI Agent工具箱

Google Workspace CLI — one command-line tool for Drive, Gmail, Calendar, Sheets, Docs, Chat, Admin, and more. Dynamically built from Google Discovery Service. I…

Independent, unofficial CLI to edit CapCut and JianYing (剪映) projects — subtitles, timing, speed, volume, templates, cut long-form to shorts. No API needed, rea…