5K
Graph Explorer

Pattern graph

See which patterns connect by shared tags, category, explicit links, or similarity. Click a node to open the pattern.

Showing 203 of 203 patterns · 0 connections · Shared tags
402-First Machine Payments (Pr...
402-First Machine Payments (Price-Before-Work Tool Purchases) Orchestration & Control emerging Server returns an HTTP 402 price quote before doing work, and the agent pays the exact amount only if it fits its budget cap
Abstracted Code Representation...
Abstracted Code Representation for Review UX & Collaboration proposed Shows reviewers pseudocode, intent summaries, and logical diffs of code changes, with drill-down to the real code to confirm the mapping
Action Caching & Replay Patter...
Action Caching & Replay Pattern Reliability & Eval emerging Records each agent action with XPath and frame metadata so later runs replay it without LLM calls, with LLM fallback when replay fails
Action-Selector Pattern
Action-Selector Pattern Orchestration & Control emerging LLM maps user intent to a pre-approved action ID with schema-validated parameters, and tool outputs never return to the selector
Adaptive Sandbox Fan-Out Contr...
Adaptive Sandbox Fan-Out Controller Reliability & Eval proposed Starts a small batch of parallel sandboxes, then scales up, stops early, or refines the prompt based on early success, variance, and error signals
Agent-Assisted Scaffolding
Agent-Assisted Scaffolding UX & Collaboration validated-in-production Agent generates initial files, boilerplate, and directory structure from a high-level description so developers start on core logic
Agent Circuit Breaker
Agent Circuit Breaker Reliability & Eval emerging Prevents agents from wasting tokens and time on repeatedly failing tools by tracking failure rates and temporarily disabling broken tool endpoints
Agent-Driven Research
Agent-Driven Research Orchestration & Control established Agent plans its own search queries, runs them across sources, reflects on gaps, and iterates until it can write a sourced report
Agent-First Tool Discovery
Agent-First Tool Discovery Tool Use & Environment emerging Build search indexes designed for agent consumers, returning structured tool metadata ranked by agent-relevant signals instead of human SEO metrics.
Agent-First Tooling and Loggin...
Agent-First Tooling and Logging Tool Use & Environment established Designs tools and logs for agent consumption with one unified log stream, structured JSON output, and agent-aware CLI flags
Agent-Friendly Workflow Design
Agent-Friendly Workflow Design UX & Collaboration emerging Gives agents clear high-level goals, room for implementation choices, structured I/O, and plan review before execution
Agent Modes by Model Personali...
Agent Modes by Model Personality Orchestration & Control emerging Offers separate working modes, each with its own prompts, tools, UI, and expectations, tuned to one model's working style
Agent-Powered Codebase Q&A / O...
Agent-Powered Codebase Q&A / Onboarding Context & Memory validated-in-production Agent indexes the codebase with embeddings and code graphs and answers natural-language questions about where code is and how it behaves
Agent Reinforcement Fine-Tunin...
Agent Reinforcement Fine-Tuning (Agent RFT) Learning & Adaptation emerging Trains model weights with reinforcement learning on real tool calls and custom graders to improve domain-specific tool use and multi-step reasoning
Agent SDK for Programmatic Con...
Agent SDK for Programmatic Control Tool Use & Environment established Exposes agent functions through an SDK and CLI so code can run the agent headless with set tools, permissions, and resource limits
Agentic Search Over Vector Emb...
Agentic Search Over Vector Embeddings Tool Use & Environment best-practice Replaces vector indexes with agent-driven grep, find, and file traversal that searches current file state on demand and refines iteratively
AI-Accelerated Learning and Sk...
AI-Accelerated Learning and Skill Development UX & Collaboration best-practice Uses AI assistants as tutors that explain errors and concepts, offer alternatives, and fade support as the developer gains skill
AI-Assisted Code Review / Veri...
AI-Assisted Code Review / Verification Feedback Loops established Uses AI tools to flag issues, summarize change intent, and explain code so human reviewers focus on intent and business logic
AI Web Search Agent Loop
AI Web Search Agent Loop Tool Use & Environment emerging Coordinator agent translates queries, spawns parallel search workers across domains and time ranges, refines iteratively, and answers with citations
Anti-Reward-Hacking Grader Des...
Anti-Reward-Hacking Grader Design Reliability & Eval emerging Design reward functions with multi-criteria evaluation and iterative hardening to prevent models from gaming graders, ensuring training rewards align with actual task quality.
Approval-Bound Transactional W...
Approval-Bound Transactional Write Reliability & Eval emerging Binds a human approval to one exact plan version, then applies the write, checks the postcondition and records the idempotency key in a single database transaction
Asynchronous Coding Agent Pipe...
Asynchronous Coding Agent Pipeline Reliability & Eval emerging Splits inference, tool execution, reward modeling, and learning into asynchronous workers linked by message queues so GPUs stay busy
Authenticated Authority Channe...
Authenticated Authority Channel Security & Safety emerging Preserve a distinguishable channel for authenticated intent so retrieved content can inform reasoning without granting itself authority.
Autonomous Workflow Agent Arch...
Autonomous Workflow Agent Architecture Orchestration & Control emerging Runs multi-step engineering workflows in containers with tmux sessions, sleep-based monitoring, checkpoints, and context-aware error recovery
Background Agent with CI Feedb...
Background Agent with CI Feedback Feedback Loops validated-in-production Runs the agent in the background on its own branch and uses CI results as feedback to patch failures until green or blocked
Black-Box Skill Invocation
Black-Box Skill Invocation Security & Safety emerging Shares skills through input and output schemas only and runs them on the provider side, so prompts and code never cross the boundary
Board-Mediated Async Inter-Age...
Board-Mediated Async Inter-Agent Coordination Orchestration & Control validated-in-production Routes agent-to-agent messages through comment threads on board cards, with a prefix that creates an inbox card to notify the target agent
Budget-Aware Model Routing wit...
Budget-Aware Model Routing with Hard Cost Caps Orchestration & Control established Routes each request to the cheapest model that meets its needs under hard cost caps, and escalates only when quality gates fail
Burn the Boats
Burn the Boats Orchestration & Control emerging Removes working but obsolete features on a hard, announced deadline so the team and users move to the new approach
Canary Rollout and Automatic R...
Canary Rollout and Automatic Rollback for Agent Policy Changes Reliability & Eval established Ships agent policy changes to a small traffic slice first, monitors guardrail metrics, and rolls back to the last stable version automatically
Capability-Escrow-Receipt
Capability-Escrow-Receipt Orchestration & Control experimental-but-awesome Binds a signed capability listing, an atomic escrow-plus-hire step, and a signed work receipt into one loop for agent-to-agent payment
Chain-of-Thought Monitoring & ...
Chain-of-Thought Monitoring & Interruption UX & Collaboration emerging Streams agent reasoning and tool calls in real time so a human can interrupt and redirect early when the approach is wrong
Classify-Then-Act for Backgrou...
Classify-Then-Act for Background Agents Orchestration & Control emerging Sorts each task into has-default, needs-intent, or out-of-scope, builds only has-default work in a sandbox, and queues it for human review
CLI-First Skill Design
CLI-First Skill Design Tool Use & Environment emerging Builds each skill as a standalone CLI with JSON output and exit codes so humans and agents use the same interface
CLI-Native Agent Orchestration
CLI-Native Agent Orchestration Tool Use & Environment established Drives agents through their non-interactive CLI mode (claude -p with JSON output and exit codes) so Makefiles, Git hooks, cron jobs, and CI can script and replay agent runs
Code Mode MCP Tool Interface I...
Code Mode MCP Tool Interface Improvement Pattern Tool Use & Environment established LLMs generate TypeScript code to orchestrate MCP tools in ephemeral V8 isolates, eliminating token-heavy round-trips and enabling efficient multi-step workflows with large measured token savings.
Code-Over-API Pattern
Code-Over-API Pattern Tool Use & Environment established Agent writes code that calls tools and filters data inside a sandbox, so only summaries and samples return to the context window
Code-Then-Execute Pattern
Code-Then-Execute Pattern Tool Use & Environment emerging LLM writes a sandboxed program or DSL script, a static taint checker verifies data flows, and an interpreter runs it in a locked sandbox
Codebase Optimization for Agen...
Codebase Optimization for Agents UX & Collaboration emerging Optimizes tooling, CLIs, tests, and docs for agents first, with one-command verify loops and machine-readable output, even if human DX regresses
Commitment Ledger with Reality...
Commitment Ledger with Reality-Gated Credit Learning & Adaptation proposed Records agent promises and credits retrieved memory only after an externally verifiable outcome settles.
Compounding Engineering Patter...
Compounding Engineering Pattern Learning & Adaptation emerging After each feature, codifies agent mistakes and learnings into CLAUDE.md, slash commands, subagents, and hooks so the next feature is easier to build
Consequence-Family Coverage Au...
Consequence-Family Coverage Audit Security & Safety emerging Audit an agent's risk policy by enumerating families of consequence and asking which rule covers each, because a hand-written risk list reliably encodes one family and stays silent on the rest
Context Budget as a Governed R...
Context Budget as a Governed Resource Context & Memory validated-in-production Measures always-loaded context per source, alerts on trend growth, and puts hard spend caps and fan-out bounds on unattended agent runs
Context-Minimization Pattern
Context-Minimization Pattern Context & Memory emerging Removes untrusted user text and tool output from context after it becomes a safe structured artifact, so later steps see only trusted data
Context Window Anxiety Managem...
Context Window Anxiety Management Context & Memory emerging Enables a large context window but caps usage lower, and adds prompts that state the token budget so the model does not rush to finish
Context Window Auto-Compaction
Context Window Auto-Compaction Context & Memory validated-in-production Catches context overflow errors, compacts and validates the transcript with capped reserve headroom, and retries the request automatically
Continuous Autonomous Task Loo...
Continuous Autonomous Task Loop Pattern Orchestration & Control established Runs a loop where subagents pick the next task from a todo file, execute it in fresh context, commit it, and back off on rate limits
CriticGPT-Style Code Review
CriticGPT-Style Code Review Reliability & Eval emerging Runs a specialized critic model over generated code to find bugs, security flaws, and quality issues, and feeds critical findings back for regeneration
Cross-Agent Lesson Sharing via...
Cross-Agent Lesson Sharing via Git Context & Memory emerging Agents write solved problems as markdown lessons in a shared Git repo and search them before debugging, with GitHub Issues as the coordination layer
Cross-Cycle Consensus Relay
Cross-Cycle Consensus Relay Orchestration & Control emerging Each loop cycle reads a structured relay document, does its work, and atomically writes back decisions, open questions, and the single next action
Cross-Domain Agent Conflict Re...
Cross-Domain Agent Conflict Resolution Orchestration & Control proposed A coordination layer that cross-references recommendations from independent domain agents, detects conflicts on shared resources, and resolves them through policy-as-code.
Cross-Protocol Agent Discovery
Cross-Protocol Agent Discovery Tool Use & Environment proposed Aggregates agent metadata across fragmented registries and protocols into a unified, protocol-agnostic discovery layer (proposed; no neutral implementation verified)
Cryptographic Governance Audit...
Cryptographic Governance Audit Trail Security & Safety emerging Middleware checks each tool call against a policy file, then signs a receipt of the action with ML-DSA and appends it to a tamper-evident chain
Curated Code Context Window
Curated Code Context Window Context & Memory validated-in-production A search subagent finds the few relevant files for a task and injects only top snippets or summaries into the main coding agent's context
Custom Sandboxed Background Ag...
Custom Sandboxed Background Agent Orchestration & Control emerging Builds an in-house background coding agent that runs in sandboxed dev environments, streams progress over WebSocket, and iterates on compiler and test feedback
Dead-Man's Switch for Schedule...
Dead-Man's Switch for Scheduled Agent Jobs Reliability & Eval validated-in-production Each scheduled job writes a success sentinel to its log, and a separate checker files a task when the sentinel is missing or stale
Declarative Multi-Agent Topolo...
Declarative Multi-Agent Topology Definition Orchestration & Control emerging Define multi-agent systems declaratively in a single topology file — agents, flows, gates, hooks, group chats — then compile to platform-specific configurations for any agentic framework.
Democratization of Tooling via...
Democratization of Tooling via Agents UX & Collaboration established Non-programmers describe a tool in natural language and iterate with an AI agent that generates and fixes the code for dashboards, scripts, or small apps
Denial Tracking & Permission E...
Denial Tracking & Permission Escalation Security & Safety proposed Track repeated tool denials and auto-escalate to blanket permission prompts or fallback strategies, preventing wasted iterations in non-interactive contexts.
Design-Time File Partition as ...
Design-Time File Partition as Concurrency Control Orchestration & Control emerging Each task declares every path it will write; overlapping tasks get a blocking dependency so only disjoint work runs at once on a shared checkout
Deterministic Grader in the Lo...
Deterministic Grader in the Loop Feedback Loops emerging Replace the LLM-as-judge in an agent's self-review loop with a deterministic scorer that returns the raw counts behind each sub-score, so the agent has something specific to fix.
Deterministic Security Scannin...
Deterministic Security Scanning Build Loop Security & Safety proposed Adds SAST, SCA, and secret scanners to the build target that the agent must run after each change, so scanner failures force the agent to fix the code
Deterministic Threat Rule Scan...
Deterministic Threat Rule Scanning Security & Safety proposed Apply deterministic regex rules as a first-pass security layer to detect known threat patterns in AI agent tool calls and skill definitions.
Deterministic Zero-LLM Orchest...
Deterministic Zero-LLM Orchestration Orchestration & Control emerging A deterministic code orchestrator splits goals, runs parallel coding agents, verifies with tests, and commits, spending LLM tokens only on the one-time goal decomposition
Dev Tooling Assumptions Reset
Dev Tooling Assumptions Reset UX & Collaboration emerging Audits dev tools for human-effort assumptions and replaces tickets, PR ceremony, and sprints with immediate agent dispatch, variations, and automated tests
Discrete Phase Separation
Discrete Phase Separation Orchestration & Control emerging Splits work into separate research, planning, and implementation conversations, each with fresh context, passing only distilled outputs between phases
Disposable Scaffolding Over Du...
Disposable Scaffolding Over Durable Features Orchestration & Control emerging Treats code around the model as temporary scaffolding: build the simplest thing that works now, mark it disposable, and remove it when models improve
Distributed Execution with Clo...
Distributed Execution with Cloud Workers Orchestration & Control emerging Runs many agent sessions in parallel on cloud workers, each in its own git worktree, with dependency-aware scheduling, merge coordination, and approval gates
Dogfooding with Rapid Iteratio...
Dogfooding with Rapid Iteration for Agent Improvement Feedback Loops best-practice The agent team uses its own agent for daily work, collects feedback in low-friction channels, and ships features internally first to validate or discard them
Dual LLM Pattern
Dual LLM Pattern Orchestration & Control emerging Splits work between a privileged LLM that calls tools but never sees untrusted data and a quarantined LLM that reads untrusted data but has no tools
Dual-Rail Message Delivery
Dual-Rail Message Delivery Orchestration & Control validated-in-production Sends every inter-agent message on two independent channels through one entry point, alerts when the rails diverge, and requires ACKs to confirm completion
Dual-Use Tool Design
Dual-Use Tool Design Tool Use & Environment best-practice Builds each tool with one interface and implementation that both humans and agents can call, with the same output, permissions, and logs
Dynamic Context Injection
Dynamic Context Injection Context & Memory established Lets users inject files, folders, or saved prompts into the agent's context mid-session with @-mentions and custom slash commands
Egress Lockdown (No-Exfiltrati...
Egress Lockdown (No-Exfiltration Channel) Tool Use & Environment established Puts the agent behind a default-deny egress firewall with allowlisted destinations so stolen data has no outbound channel
Episodic Memory Retrieval & In...
Episodic Memory Retrieval & Injection Context & Memory established Writes a structured memory record after each episode and injects the top-k similar past memories as hints when a new task starts
Evidence-Layered Evaluation fo...
Evidence-Layered Evaluation for Interactive Agents Reliability & Eval emerging Scores interactive agent runs on a task outcome assertion and links each result to layered evidence: actions, screenshots, replays, network, and messages
Evidence Questions, Not Verdic...
Evidence Questions, Not Verdict Questions Security & Safety emerging Never ask the model for allow/ask/block. Ask it narrow, typed, observational questions and let code reduce the answers to the verdict, because the collapsed question is the low-confidence one.
Exact-Action Authorization Bin...
Exact-Action Authorization Binding Security & Safety proposed Bind a short-lived approval to the complete action presented to the reviewer, then rederive and compare that identity at the acting boundary before execution.
Explicit Posterior-Sampling Pl...
Explicit Posterior-Sampling Planner Orchestration & Control emerging Embeds Posterior Sampling for RL in the LLM's reasoning: sample a task model, plan, act, observe reward, and update the posterior
Extended Coherence Work Sessio...
Extended Coherence Work Sessions Reliability & Eval rapidly-improving Combines models with long coherence windows with context compaction, prompt caching, and persisted state so agents stay on task for hours
External Credential Sync
External Credential Sync Security & Safety established Reads AI credentials from other tools' keychains and config files and syncs them into the agent's store, with TTL-cached reads, refresh fences, and persisted ownership
Factory over Assistant
Factory over Assistant Orchestration & Control validated-in-production Spawns several autonomous agents in parallel with automated feedback loops such as tests and builds, and checks on them later instead of watching one agent
Failover-Aware Model Fallback
Failover-Aware Model Fallback Reliability & Eval validated-in-production Classifies each model failure by reason and falls back along a model chain only for retryable errors, failing fast on auth, billing, and user aborts
Fair-Share Task Queues with He...
Fair-Share Task Queues with Heartbeat Leases Orchestration & Control proposed Distributes tasks across a shared agent fleet using weighted deficit-round-robin scheduling, with heartbeat leases and dead-letter retries to recover from crashed workers
Feature List as Immutable Cont...
Feature List as Immutable Contract Orchestration & Control emerging Defines every feature up front in a JSON list with acceptance steps; the agent may only flip a feature to passing after it verifies it
Filesystem-Based Agent State
Filesystem-Based Agent State Context & Memory established Agents persist intermediate results and working state to files, creating durable checkpoints that enable workflow resumption, recovery from failures, and support for long-running tasks.
Filesystem-Mediated Host Deleg...
Filesystem-Mediated Host Delegation Tool Use & Environment emerging Lets a sandboxed agent write request files to a shared spool directory that a host daemon runs against a whitelist, with idempotent results the agent polls for
Frontier-Focused Development
Frontier-Focused Development Learning & Adaptation emerging Targets only state-of-the-art models, picks the best model per use case with no user model selector, and expects to rework the product every few months
Graph of Thoughts (GoT)
Graph of Thoughts (GoT) Feedback Loops emerging Represents reasoning as a directed graph of thoughts so the model can branch, aggregate, refine, and revisit reasoning paths
Hook-Based Safety Guard Rails ...
Hook-Based Safety Guard Rails for Autonomous Code Agents Security & Safety validated-in-production Runs small shell scripts on PreToolUse and PostToolUse hooks to block destructive commands, lint edits, and warn on context use outside the agent's reasoning
Human-in-the-Loop Approval Fra...
Human-in-the-Loop Approval Framework UX & Collaboration validated-in-production Systematically insert human approval gates for designated high-risk functions while maintaining agent autonomy for safe operations, with multi-channel approval interfaces and comprehensive audit trails.
Hybrid LLM/Code Workflow Coord...
Hybrid LLM/Code Workflow Coordinator Orchestration & Control proposed Lets each workflow pick an LLM or a code script as its coordinator, so teams prototype with the LLM and move to reviewed code when determinism matters
Incident-to-Eval Synthesis
Incident-to-Eval Synthesis Feedback Loops emerging Converts each production incident into executable eval cases with pass/fail criteria and gates future releases on them
Inference-Healed Code Review R...
Inference-Healed Code Review Reward Feedback Loops proposed Replaces a binary tests-passed reward with a critic that scores correctness, style, performance, and security, explains low scores, and combines them
Inference-Time Scaling
Inference-Time Scaling Orchestration & Control established Spends extra compute at inference time with best-of-N sampling, longer reasoning, self-refinement, search, and verification to improve output quality
Initializer-Maintainer Dual Ag...
Initializer-Maintainer Dual Agent Architecture Orchestration & Control emerging Uses a one-time initializer agent to create the feature list, progress files, and bootstrap script, then a coding agent that resumes from them one feature per session
Intelligent Bash Tool Executio...
Intelligent Bash Tool Execution Tool Use & Environment validated-in-production Runs agent shell commands through a multi-mode executor that uses a PTY when needed, falls back to direct exec, and manages approvals, background jobs, and signals
Inversion of Control
Inversion of Control Orchestration & Control validated-in-production Gives the agent tools, a high-level objective, and guardrails, then lets it choose sequencing and recovery while humans set policy and review
Isolated VM per RL Rollout
Isolated VM per RL Rollout Security & Safety emerging Spin up an isolated virtual machine for each RL rollout to prevent cross-contamination between parallel agent executions, ensuring safe training with destructive tool access.
Iterative Multi-Agent Brainsto...
Iterative Multi-Agent Brainstorming Orchestration & Control experimental-but-awesome Spawns several agents in parallel on the same problem, often with different perspectives, then merges their ideas into one set of options
Iterative Prompt & Skill Refin...
Iterative Prompt & Skill Refinement Feedback Loops emerging Combines a feedback channel, editable prompt documents, log-driven skill fixes, and usage dashboards to improve agent prompts and skills continuously
Judge-Gated Finding Exchange B...
Judge-Gated Finding Exchange Between Parallel Sub-Agents Orchestration & Control emerging Puts an append-only board between parallel sub-agents and lets a small judge decide what goes on it, who receives each note, and when a later note retracts an earlier one
Lane-Based Execution Queueing
Lane-Based Execution Queueing Orchestration & Control validated-in-production Routes work into named lanes, each with its own queue and concurrency limit, so sessions never interleave and background jobs never block users
Language Agent Tree Search (LA...
Language Agent Tree Search (LATS) Orchestration & Control emerging Runs Monte Carlo Tree Search over reasoning steps, with the LLM generating candidate actions and scoring partial solutions to pick the best path
Latent Demand Product Discover...
Latent Demand Product Discovery UX & Collaboration best-practice Builds hackable, extensible products, watches how power users repurpose them, and turns the most frequent workarounds into supported features
Layered Configuration Context
Layered Configuration Context Context & Memory established Loads context files from enterprise, user, project, and local levels automatically and merges them into the agent's baseline context
Lethal Trifecta Threat Model
Lethal Trifecta Threat Model Security & Safety best-practice Classifies each tool by private-data access, untrusted-content exposure, and external communication, and blocks any execution path that has all three
LLM-Friendly API Design
LLM-Friendly API Design Tool Use & Environment emerging Designs APIs with visible versions, self-descriptive names and schemas, simple calls, actionable errors, and few indirection levels so LLMs call them correctly
LLM Map-Reduce Pattern
LLM Map-Reduce Pattern Orchestration & Control emerging Processes each untrusted document in its own sandboxed LLM call with a constrained output, then aggregates the results with deterministic code
LLM Observability
LLM Observability Reliability & Eval emerging Sends agent runs to an LLM observability platform for span-level traces of each LLM call and tool use, plus aggregate cost, latency, and success metrics
Local-First Credential Broker
Local-First Credential Broker Security & Safety emerging Keep raw secrets out of the agent process by injecting credentials at the network layer through a local broker, rather than handing the agent environment variables or config files.
Markdown Polis — Multi-Vendor ...
Markdown Polis — Multi-Vendor Agent Coordination via Filesystem Constitution Orchestration & Control emerging Coordinates agents from different vendors through versioned markdown files: capability cards, work contracts, and bandit routing that learns from settled work
MCP Pattern Injection
MCP Pattern Injection Tool Use & Environment emerging Runs an MCP server that exposes framework best-practice patterns as tools, so the coding assistant fetches current patterns on demand
Memory Reinforcement Learning ...
Memory Reinforcement Learning (MemRL) Learning & Adaptation proposed Stores memories with learned utility scores, retrieves by similarity then re-ranks by utility, and updates scores from outcomes while the LLM stays frozen
Memory Synthesis from Executio...
Memory Synthesis from Execution Logs Context & Memory emerging Has the agent write a structured diary per task, then runs synthesis agents over many diaries to turn recurring patterns into rules, commands, and tests
Merged Code + Language Skill M...
Merged Code + Language Skill Model Learning & Adaptation emerging Fine-tunes separate language and code specialists from the same base model, then merges their weights into one model instead of one large joint training run
Milestone Escrow for Agent Res...
Milestone Escrow for Agent Resource Funding UX & Collaboration proposed Holds agent funding in escrow and releases each payment only after independent verification of a measurable milestone
Multi-Model Orchestration for ...
Multi-Model Orchestration for Complex Edits Orchestration & Control validated-in-production Splits complex code edits across specialized models: a retrieval model gathers context, a large model writes the changes, and smaller models apply them
Multi-Platform Communication A...
Multi-Platform Communication Aggregation Tool Use & Environment emerging Queries every communication platform in parallel through adapters that share one schema, then merges, deduplicates, and ranks the results
Multi-Platform Webhook Trigger...
Multi-Platform Webhook Triggers Tool Use & Environment emerging Starts agent workflows from Slack, Notion, and Jira webhooks, emoji reactions, and schedules, with idempotency and signature checks on each event
Artifact-Driven Analysis Pipel...
Artifact-Driven Analysis Pipeline Orchestration Orchestration & Control emerging Has an agent run independent analysis scripts, read their structured reports, and merge them into one final report or visualization
No-Token-Limit Magic
No-Token-Limit Magic Reliability & Eval experimental-but-awesome Removes hard token limits during prototyping to learn what good behavior needs, then compresses context only after quality is stable and measured
Non-Custodial Spending Control...
Non-Custodial Spending Controls Security & Safety emerging Puts a policy layer between the agent and the transaction signer that checks each intent against allowlists, budgets, and rate limits, and fails closed
Non-Generative Judgment Routin...
Non-Generative Judgment Routing with Typed Escalation Orchestration & Control emerging Sends decision-only loop steps as batched typed questions to a non-generative judgment model and escalates what it cannot answer back to the LLM
One OS User per Agent
One OS User per Agent Security & Safety validated-in-production Give each long-lived agent its own OS user account and a templated init-system unit, so identity, supervision and privilege come from the host instead of from a per-agent container.
Opponent Processor / Multi-Age...
Opponent Processor / Multi-Agent Debate Pattern Orchestration & Control emerging Spawns agents with opposing roles on the same context, lets them critique each other, then synthesizes their positions to expose bias and blind spots
Oracle and Worker Multi-Model ...
Oracle and Worker Multi-Model Approach Orchestration & Control emerging Uses a fast, low-cost worker model for most tool use and code generation, and lets it consult an expensive oracle model when it is stuck
Orchestration Prompt-Writing B...
Orchestration Prompt-Writing Benchmark Reliability & Eval emerging Scores an orchestrator on whether its sub-agent prompts give the right information fragments to the right roles, separate from end-task success
Out-of-Process Provider-Bounda...
Out-of-Process Provider-Boundary Replay Reliability & Eval emerging Capture an agent run at the model-provider HTTP boundary from outside the process, then serve those bytes back so the agent re-executes its own logic with no provider call.
Output Verification Loop
Output Verification Loop Reliability & Eval emerging Verify LLM outputs by extracting individual claims, checking each against evidence sources, and returning per-claim trust scores before acting on the result.
Own-Check Fault Injection
Own-Check Fault Injection Reliability & Eval emerging Plant a controlled fault inside a running agent pipeline and score whether the pipeline's own checks emit a detection act, keeping detected, reacted and recovered as separate verdicts.
Parallel Tool Call Learning
Parallel Tool Call Learning Orchestration & Control emerging Uses agent reinforcement fine-tuning to teach the model to issue independent tool calls in parallel, which cuts sequential rounds and latency
Conditional Parallel Tool Exec...
Conditional Parallel Tool Execution Orchestration & Control validated-in-production Runs a batch of tool calls in parallel when all are read-only and in sequence when any modifies state, then returns results in request order
Patch Steering via Prompted To...
Patch Steering via Prompted Tool Selection Tool Use & Environment emerging Tells the agent in the prompt which patch tool or edit format to use, with usage examples, negative rules, and a fallback order
Persistent Test Memory Feedbac...
Persistent Test Memory Feedback Loop Learning & Adaptation emerging Preserve validated test lessons so an agent can reuse successful paths, recognize recurring failures, and retire stale knowledge.
PII Tokenization
PII Tokenization Security & Safety established Replaces PII in tool results with placeholder tokens before the model sees them and restores the real values in outgoing tool calls
Plan-Then-Execute Pattern
Plan-Then-Execute Pattern Orchestration & Control established Has the LLM fix the full sequence of tool calls before it reads untrusted data, then runs that sequence so tool outputs change only parameters
Planner-Worker Separation for ...
Planner-Worker Separation for Long-Running Agents Orchestration & Control emerging Splits agents into planners that create tasks, workers that complete them in isolation, and a judge that decides each cycle whether to continue
Policy-Gated Tool Proxy
Policy-Gated Tool Proxy Security & Safety emerging Insert a transparent proxy between agents and tool servers that evaluates every tool call against a policy engine before forwarding, producing an immutable audit trail of all decisions.
Precomputed Code Graph Lookup
Precomputed Code Graph Lookup Tool Use & Environment emerging Answer an agent's structural code questions from a precomputed, deterministic graph queried by intent-shaped verbs, instead of a retrieval-time search-and-read loop.
Proactive Agent State External...
Proactive Agent State Externalization Context & Memory emerging Gives agents note templates, completeness checks, and an external memory fallback so self-written notes keep objectives, decisions, and knowledge gaps
Proactive Trigger Vocabulary
Proactive Trigger Vocabulary UX & Collaboration emerging Gives each skill an explicit, documented list of trigger phrases and patterns so input routes to skills predictably, with optional proactive activation
Progressive Autonomy with Mode...
Progressive Autonomy with Model Evolution Orchestration & Control best-practice Audits prompts and orchestration after each model upgrade and removes the scaffolding that evals show the new model no longer needs
Progressive Complexity Escalat...
Progressive Complexity Escalation Orchestration & Control emerging Deploys agents on low-complexity, high-reliability tasks first and unlocks higher capability tiers when metrics and human review gates show proven reliability
Progressive Disclosure for Lar...
Progressive Disclosure for Large Files Context & Memory emerging Puts only file metadata in the prompt and gives the agent load, peek, and extract tools to pull file content into context on demand
Progressive Tool Discovery
Progressive Tool Discovery Tool Use & Environment established Organizes tools in a browsable hierarchy and lets the agent load names, descriptions, or full schemas only for the tools it needs
Prompt Caching via Exact Prefi...
Prompt Caching via Exact Prefix Preservation Context & Memory emerging Keeps static prompt content first in a fixed order and only appends new messages, including config changes, so each request reuses the cached prefix
Reasoning-Token Firewall
Reasoning-Token Firewall Reliability & Eval emerging Assemble an agent's result only from answer-typed stream events, never by string-stripping interleaved reasoning tokens.
Recursive Best-of-N Delegation
Recursive Best-of-N Delegation Orchestration & Control emerging Runs several parallel candidate workers per subtask in a recursive agent tree, scores them with tests and a judge, and promotes the best result upward
Reflection Loop
Reflection Loop Feedback Loops established Scores each draft against a fixed rubric, feeds the critique into a revision, and repeats until the draft passes a threshold or the retry budget ends
Rendered UI Finish Gate
Rendered UI Finish Gate Feedback Loops emerging Verify an agent-built interface against its intended design contract, required states, interaction semantics, and rendered output before merging.
Rich Feedback Loops > Perfect ...
Rich Feedback Loops > Perfect Prompts Feedback Loops validated-in-production Returns compiler errors, test failures, lint output, and human feedback to the agent after each tool call so it can plan fixes and self-correct
RLAIF (Reinforcement Learning ...
RLAIF (Reinforcement Learning from AI Feedback) Reliability & Eval established Uses an AI model guided by written principles to critique outputs and label preferences, which train a reward model that optimizes the policy
Sandboxed Tool Authorization
Sandboxed Tool Authorization Security & Safety validated-in-production Filters an agent's tools through deny-first allow and deny patterns, profile presets, and subagent policies that inherit parent restrictions
Schema-Guided Graph Retrieval ...
Schema-Guided Graph Retrieval for Multi-Hop Reasoning Context & Memory emerging Use one shared domain schema to align graph construction, schema evolution, query decomposition, and typed retrieval so multi-hop reasoning over private knowledge stays precise as domains change.
Schema Validation Retry with C...
Schema Validation Retry with Cross-Step Learning Reliability & Eval emerging Retries failed structured outputs with the validation errors as feedback and adds recent errors from earlier steps to later prompts so mistakes do not repeat
Seamless Background-to-Foregro...
Seamless Background-to-Foreground Handoff UX & Collaboration emerging Lets a user take over a background agent's unfinished work in the foreground, with the agent's branch, PR, and summaries carried over as context
Self-Critique Evaluator Loop
Self-Critique Evaluator Loop Feedback Loops established Trains a judge model on its own synthetic comparisons of candidate outputs and uses it as a reward model or quality gate for the main agent
Self-Discover: LLM Self-Compos...
Self-Discover: LLM Self-Composed Reasoning Structures Learning & Adaptation emerging Has the LLM select, adapt, and compose reasoning modules into a task-specific reasoning structure, then solve the task by following that structure
Self-Identity Accumulation
Self-Identity Accumulation Context & Memory emerging Injects a persistent identity document at session start and updates it with new user insights at session end through lifecycle hooks
Self-Rewriting Meta-Prompt Loo...
Self-Rewriting Meta-Prompt Loop Orchestration & Control emerging Has the agent reflect after each episode, draft edits to its own system prompt, validate them through guardrails, and save the new version
Semantic Context Filtering Pat...
Semantic Context Filtering Pattern Context & Memory emerging Extracts only the semantic or interactive parts of raw data, such as accessibility trees or relevant fields, before sending it to the LLM
Session-Scoped Context Runtime...
Session-Scoped Context Runtime for Agent Tools Context & Memory emerging Interpose a context runtime that caches structured reads and normalizes tool output so sessions reuse compact representations instead of repeating raw tokens.
Shell Command Contextualizatio...
Shell Command Contextualization UX & Collaboration established Lets the user run a shell command with a prefix such as ! and injects the command and its full output into the agent's context
Shipping as Research
Shipping as Research Learning & Adaptation emerging Releases reversible, instrumented features to learn whether they work, then doubles down or removes them based on usage data and feedback
Signal-Driven Agent Activation
Signal-Driven Agent Activation Orchestration & Control emerging Watches external sources for structured signals and starts predefined agent workflows when declarative rules with thresholds and cooldowns match
Skill Activation as a Precisio...
Skill Activation as a Precision/Recall Measurement Reliability & Eval emerging Measures skill routing as classification with a labelled prompt dataset, per-skill precision and recall, and CI thresholds that fail the build
Skill Library Evolution
Skill Library Evolution Learning & Adaptation established Saves working agent code as reusable skills in a skills directory, documents and tests them over time, and loads skill details only on demand
Soulbound Identity Verificatio...
Soulbound Identity Verification Security & Safety emerging Binds agent identity to a non-transferable credential with a committed state hash and logs signed state changes so verifiers can check continuity
Spec-As-Test Feedback Loop
Spec-As-Test Feedback Loop Feedback Loops emerging Generates executable tests from the spec on every spec or code commit and opens agent PRs that fix code or flag unclear spec parts
Specification-Driven Agent Dev...
Specification-Driven Agent Development Orchestration & Control emerging Makes a version-controlled spec file the agent's main input, scaffolds code from it, and links each artifact back to a spec clause
Spectrum of Control / Blended ...
Spectrum of Control / Blended Initiative UX & Collaboration validated-in-production Gives users several autonomy modes, from inline completion to background agents, and lets them switch modes per task
Static Service Manifest for Ag...
Static Service Manifest for Agents Tool Use & Environment emerging Serves a static llms.txt or agent.json file at a well-known URL that lists services, auth, and limits so agents can plan before they call
Stop Hook Auto-Continue Patter...
Stop Hook Auto-Continue Pattern Orchestration & Control emerging Runs a stop hook after each agent turn that checks success criteria and makes the agent continue until tests or checks pass
Structured Output Specificatio...
Structured Output Specification Reliability & Eval established Constrain agent outputs using deterministic schemas that enforce structured, machine-readable results, enabling reliable validation, parsing, and integration with downstream systems.
Sub-Agent Spawning
Sub-Agent Spawning Orchestration & Control validated-in-production Lets the main agent spawn sub-agents with fresh context and scoped tools to work on subtasks in parallel, then merges their results
Subagent Compilation Checker
Subagent Compilation Checker Reliability & Eval emerging Spawns one subagent per module to build and check it, and returns only a short structured error list or artifact reference to the main agent
Subject Hygiene for Task Deleg...
Subject Hygiene for Task Delegation Orchestration & Control emerging Requires a specific action-plus-target subject on every subagent task so each delegated job stays traceable and easy to reference
Surface-Scoped System Prompt
Surface-Scoped System Prompt Context & Memory established Ships one system prompt per invocation surface and selects it at launch, so an interactive session, a headless run and a task mode each get instructions written for that surface
Swarm Migration Pattern
Swarm Migration Pattern Orchestration & Control validated-in-production Main agent orchestrates 10+ parallel subagents working simultaneously on independent migration chunks for large-scale framework upgrades, lint rule rollouts, and API migrations.
Team-Shared Agent Configuratio...
Team-Shared Agent Configuration as Code UX & Collaboration best-practice Check agent configuration into version control as code, enabling consistent behavior across teams, faster onboarding, and collaborative improvement through PRs and code review.
Three-Stage Perception Archite...
Three-Stage Perception Architecture Orchestration & Control established Splits the agent into separate perception, processing, and action stages so each stage can be built, tested, and scaled on its own
Tier Auto-Apply by Mechanical ...
Tier Auto-Apply by Mechanical Impact Reliability & Eval emerging Decides which self-proposed changes auto-apply by matching the real diff's file paths and change kinds against a trusted tier table, not by finding text
Tool Capability Compartmentali...
Tool Capability Compartmentalization Orchestration & Control emerging Splits tools into reader, processor, and writer classes with scoped permissions and blocks tool chains that combine private data, untrusted input, and external writes
Tool Search Lazy Loading
Tool Search Lazy Loading Context & Memory established Dynamically load tools via search instead of preloading all available tools to reduce context usage
Tool Selection Guide
Tool Selection Guide Orchestration & Control emerging Maps each task type to a preferred tool: Glob, Grep, and Read to explore, Edit to modify, Bash to verify, and Task with a clear subject to delegate
Tool Use Incentivization via R...
Tool Use Incentivization via Reward Shaping Feedback Loops emerging Gives dense RL rewards for useful intermediate tool calls such as compile, lint, and test so the agent learns to use tools instead of only thinking
Tool Use Steering via Promptin...
Tool Use Steering via Prompting Tool Use & Environment best-practice Tells the agent in the prompt which tool to use, how to learn a custom tool, and which shorthands map to tool sequences
Tracker-as-Desired-State Recon...
Tracker-as-Desired-State Reconciliation Orchestration & Control emerging Derive agent dispatch by re-reading workflow states in the team's issue tracker every tick, keeping only a short-lived claim locally, so missed events, restarts, and human edits all self-heal.
Transitive Vouch-Chain Trust
Transitive Vouch-Chain Trust Security & Safety emerging Builds a graph of signed vouches between agents and derives trust in an unknown agent from the chain back to a trusted one, with decay at each hop
Tree-of-Thought Reasoning
Tree-of-Thought Reasoning Orchestration & Control established Expands a tree of candidate reasoning steps, scores partial states, prunes weak branches, and picks the best path instead of one linear chain
Unified Tool Gateway
Unified Tool Gateway Tool Use & Environment emerging Route all agent tool calls through a single gateway that handles discovery, authentication, billing, and execution across many heterogeneous providers.
Variance-Based RL Sample Selec...
Variance-Based RL Sample Selection Learning & Adaptation emerging Runs the base model several times per sample and trains RL only on samples with score variance, skipping ones that are always right or always wrong
Verbose Reasoning Transparency
Verbose Reasoning Transparency UX & Collaboration best-practice Lets users open a verbose view on demand that shows the agent's interpretation, tool choices, intermediate steps, and raw tool outputs
Versioned Constitution Governa...
Versioned Constitution Governance Reliability & Eval emerging Stores the agent constitution in a signed Git repository where the agent can only propose changes and reviewers or CI gates merge them
Virtual Machine Operator Agent
Virtual Machine Operator Agent Tool Use & Environment established Gives the agent a dedicated virtual machine where it can run code, install packages, use the file system, and operate CLI tools
Visual AI Multimodal Integrati...
Visual AI Multimodal Integration Tool Use & Environment established Adds multimodal models to the agent so it can analyze images, video, and screenshots and combine them with text to reason and act
Reliability Problem Map Checkl...
Reliability Problem Map Checklist for RAG and Agents Reliability & Eval proposed Runs a fixed 16-question failure checklist on each RAG or agent incident, maps the result to repair actions, and re-tests the same failing case
Workflow Evals with Mocked Too...
Workflow Evals with Mocked Tools Reliability & Eval established Runs complete agent workflows against mocked tools in CI and checks which tools were called plus agent-as-judge quality criteria
Working Memory via Task Tracki...
Working Memory via Task Tracking Context & Memory emerging Keeps an explicit task list with status, blockers, and next steps during the session so agent and user can track progress
Workspace-Native Multi-Agent O...
Workspace-Native Multi-Agent Orchestration Orchestration & Control emerging Runs agents inside the team's workspace platform so they share its memory, knowledge sources, event triggers, and integrations with humans
Zero-Knowledge Verified Agent ...
Zero-Knowledge Verified Agent Egress Security & Safety emerging Intercepts each outbound HTTP or MCP call, proves its claims against a private source of truth with a zero-knowledge proof, and blocks calls that fail
Zero-Trust Agent Mesh
Zero-Trust Agent Mesh Security & Safety emerging Gives each agent a cryptographic identity and verifies identity, signed delegation tokens, and chain depth on every inter-agent request