Results for “agent-loop”

Automate multi-turn adversarial conversations against LLM agents using Microsoft PyRIT, including Crescendo and Tree-of-Attacks-with-Pruning (TAP) attack chains with scorer feedback loops.

Instrument a shipped product's AI agents with tracing and observability so you can see what they did, why outputs happened, and what each run cost. Plain-language primer plus free-tier-first backend selection (Langfuse, Phoenix, LangSmith, Braintrust) and OpenTelemetry/OpenInference instrumentation. Load when the user asks to add observability, add tracing, instrument my agents, see what my agent is doing in production, set up Langfuse or Phoenix or LangSmith, debug why my agent gave a bad answer, or track LLM cost per request. Also fires when agent-system-architecture or setup-evaluation requires an observability plan for an agent-chain product. NOT for tracing the coding agent itself — that is run-trace. Precondition for runtime-learning-loop.

ScepticAgent internal architecture reference. Use whenever the user asks how the extension works, wants to add a new AI provider, add a new highlight category, debug communication between components, understand the agent loop or streaming, or work with provider routing and the Gemini CORS proxy.

Run the Ouroboros specification-first development loop: reduce ambiguity with a Socratic interview grounded in live git data (commits, churn, contributors), freeze an immutable seed/spec, render the execution plan through spec-kit (/speckit.plan → /speckit.tasks), execute against that contract through cli-anything agent-native CLI harnesses (cli-hub, --json evidence), verify before claiming success, and keep looping until completion is actually verified. Use when the user wants spec-first clarification, git-aware interviews, immutable requirements, drift-aware implementation, harness-driven execution, or a persistent completion loop that should keep going until tests / checks / acceptance criteria pass. Triggers on: ooo, ouroboros, interview, seed, run workflow, evaluate, evolve, ooo ralph, specification first, socratic interview, git-aware interview, ambiguity reduction, execution plan, cli harness execute, persistent completion.

Routes a research topic through a structured two-phase workflow: generate an extensible outline, then fan out parallel web-search agents to investigate each item into validated JSON, producing a complete markdown report with table of contents.

One-shot setup and operation guide for the solana-clawd agentic engine. Use when: cloning the repo, setting up MCP tools, starting the Telegram bot, deploying to Fly.io/Netlify, hatching blockchain buddies, running OODA loops, configuring voice mode (ElevenLabs + Grok), minting Metaplex agents, managing the vault, running the worker swarm, or contributing to the project. Covers all 31 MCP tools, 18 buddy species, 9 spinners, 60+ Telegram commands, 95 skills, and the full repo structure.

Sid Orchestra — portable multi-agent swarm for any Cursor workspace. Run IDs, lock leases, plan critic, canary harness, PASS/FAIL evals, anti-hallucination. Use when the user says sid orchestra, @sid-orchestra, sid swarm, sid evals, or wants research→plan→build→review with a bus and loop. Available globally from ~/.cursor/skills.

Run a structured retrospective after development-phase runs of your product's agents — interview the owner in plain language about what went well and poorly, draft ranked improvement hypotheses, then design and run small n=1/n=2 experiments with pre-declared success criteria, guardrails, stop conditions, and a cost/ROI kill-switch. Load when the user says how did that run go, retro this run, the agent output was bad, what should we improve, draft hypotheses, run a small experiment, or after repeated dev runs of an agentic system produce uneven quality. Priority: output quality over performance over cost, each with diminishing-returns stops. NOT a product A/B test (experimentation), NOT coding-agent harness repair (harness-evolution), NOT production-scale learning (runtime-learning-loop).

This skill should be used when the user wants to implement features or fix bugs using test-driven development. Enforces the RED-GREEN-REFACTOR cycle with vertical slicing, context isolation between test writing and implementation, human checkpoints, and auto-test feedback loops. Uses multi-agent orchestration with the Task tool for architecturally enforced context isolation. Supports Jest, Vitest, pytest, Go test, cargo test, PHPUnit, and RSpec.

Turn a weak/cheap "Flash" model into a "Pro" performer by wrapping HKUDS UpSkill — captures agent session failures, has a strong Teacher model analyze them and draft a skill, then validates it against the weak Student model in a closed Ralph Loop (up to 3 rounds) before storing it for automatic reuse. Use when the user wants to install UpSkill, run `/upskill-init`, `/upskill-configure`, `/upskill-build`, `/upskill-run`, `/upskill-list`, `/upskill-status`, `/upskill-mode`, `/upskill-model`, `/upskill-remove`, or `/upskill-uninstall`, wants a cheap model to perform closer to a Pro model without switching, or wants a good session (success or failure) distilled into a validated skill. Triggers on: upskill, up-skill, flash to pro, teacher student distillation, ralph loop skill validation, distill agent failures into skills. Routes skill-quality ratcheting to `skill-autoresearch`, scaffolding to `write-a-skill`, and spec-compliance rewrites to `skill-standardization`.

Use Orca orchestration for structured multi-agent coordination: threaded messages, blocking ask/reply flows, task dispatch, worker_done/escalation waits, task DAGs, decision gates, coordinator loops, or decomposing work across agents. Use `orca-cli` instead for full ownership handoffs, including requests phrased as "hand off", "handoff", "handover", "give this to another agent", or "another worktree" when the user did not explicitly ask to supervise, monitor, wait for results, or coordinate a DAG. Use `orca-cli` for ordinary terminal control, lightweight terminal prompts, shell commands, Orca worktree management, reading or waiting on terminals, and automation of the browser embedded inside Orca. Use Computer Use for browser windows, webviews, Orca app UI, or desktop UI outside Orca's embedded browser.

Bluesky + X social loop. The bundled notifications poller runs `social-cli sync` on cron (default `*/15`), parses the per-platform `inbox-<platform>.yaml` files, and wakes the agent in batches of up to 3 never-seen notifications per turn. The optional feed poller runs `social-cli feed` every 2h for timeline scanning. Agent reads inbox, writes `outbox-<platform>.yaml`, runs `social-cli dispatch`. Also supports one-shot commands (post/reply/thread/like). Opt-in: install the skill, drop `.env` credentials into `<home>/state/pollers/social-cli-notifications/`. Companion to the `pollers` framework skill and the `world-scanning` skill.

Complete SolanaOS agent skill — install, configure, and operate the autonomous Solana trading runtime with Honcho v3 epistemological memory, multi-venue perp trading (Hyperliquid + Aster), on-chain intelligence with USD pricing, Telegram bot, gateway API, Tailscale mesh, hardware integration, and cross-session recall. Use when asked to install SolanaOS, query Solana blockchain data, manage wallets, run OODA trading loops, configure strategies, control BitAxe mining fleets, pair Seeker devices, or operate any SolanaOS runtime surface.

Two-party working posture — user as director, agent as executor. Every fork, tradeoff, and taste choice is surfaced via batched AskUserQuestion with structural framing, a recommended default, and concrete previews when comparison is visual, so the human steers direction while the agent handles implementation. Eliminates the review-bottleneck (no giant diff to approve at the end — review is distributed across picks) and prevents codebase-understanding debt (the user remembers the architecture because they picked it). Use whenever the user invokes /duet, or says "work with me", "ask before", "check with me", "I want to decide", "don't assume", "human-in-the-loop", "co-author", "pair with me", "duet", or whenever a task clearly involves aesthetic, architectural, or irreversible strategic decisions — even without those exact words. Pair with the Duet output style to minimize cognitive load between picks.

Use this skill whenever a developer wants to turn an idea into a complete, production-ready, end-to-end system plan BEFORE writing any code. Trigger on 'plan this system', 'design the architecture for', 'help me blueprint', 'deep plan for X', 'break this idea into components', 'expand into action points', 'full implementation plan', or when the user pastes a project idea wanting architecture, components, pipelines, and file-level execution mapped out. Casual phrasing also triggers: 'help me think this through end-to-end', 'plan before coding'. Also covers living-workspace patterns: self-improving knowledge bases, reflection loops with auditor agents, four-agent teams, schema-as-code, wiki health scoring. **Resume triggers**: 'where did we leave off', 'continue this plan', 'resume my blueprint' - rehydrates state from disk via SESSION.md/NEXT.md/decisions/. Web research is mandatory every invocation.

Expert strategy advisor for products built on LLMs or agents — not general product strategy (see `product-strategy` for that). Use this — proactively and without waiting to be asked — whenever choosing where to apply AI in a product, deciding between RAG and fine-tuning, designing how much autonomy an AI feature should have, evaluating whether an AI feature is actually defensible, or deciding whether to add AI to a feature at all. Also triggers for: "should this be an agent or a simple LLM call", "how much autonomy should this feature have", "RAG vs fine-tuning", "is this AI feature defensible", "our AI feature keeps hallucinating and users don't trust it", "should we build this with AI or just ship it deterministic", "AI product wedge", "what happens to this feature when the models get better", "human-in-the-loop design for AI features". Produces a decision-focused brief: the wedge, the architecture choice, the autonomy level, and the defensibility bet — each with an explicit trade-off.