Agent
The LLM escalation tail of the Phase Recovery Pipeline (ADR-0044). Dispatched ONLY for a terminal state the deterministic FatalPaneDetector could not classify (CauseUnknown) — reads the incident + recent pane tail, classifies it into the typed cause vocabulary (modelinvalid | cliselfupdated | deadshell), extracts the…
Agent
Fault localization agent for the Evolve Loop (Plan archetype). The advisor INSERTS this phase on bugfix cycles before build to identify the files and lines that contain the bug.
Agent
Non-determinism gate for the Evolve Loop (Evaluate archetype). The advisor INSERTS this phase after Build whenever the cycle touched at least one file (build.filestouched > 0) to re-run the affected tests under -count=N and -shuffle and prove their verdicts are stable.
Agent
Forces analysis and industry attractiveness evaluation agent for the Evolve Loop (Evaluate archetype). Evaluates competitive rivalry, buyer/supplier power, entry/substitute threats, and delivers an attractiveness verdict.
Agent
Adversarial frontend design review agent for the Evolve Loop (Evaluate archetype). The advisor INSERTS this phase after Build whenever scout.goaltype == "frontend-ui", to judge changed UI for production-grade design quality and BLOCK on design-system violations or broken responsive states.
Agent
Fuzz testing runner and evaluator (Evaluate archetype).
Agent
Delivery-semantics auditor for the Evolve Loop (Evaluate archetype). The advisor INSERTS this phase after Build on messaging cycles (scout.goaltype == "messaging") to prove that every changed message/event handler is safe to process the SAME message twice under at-least-once delivery — and BLOCKS when a handler…
Agent
Structured 4-section debrief to prevent incident recurrence when root causes go unrecorded.
Agent
Pre-Scout intent capture agent. Takes a vague user goal, classifies it via the Ask-when-Needed framework, and emits a structured intent.md with goal/non-goals/constraints/interfaces/acceptance/assumptions/challenged-premises/risk-level. Mandatory ≥1 challenged premise.
Agent
Internationalization audit agent for the Evolve Loop (Evaluate archetype). The advisor INSERTS this phase after Build whenever the cycle's scout.goaltype == "i18n" to verify changed UI/string code externalizes copy and formats data locale-aware before launch.
Agent
Market sizing and feasibility analysis agent for the Evolve Loop (Evaluate archetype). Estimates TAM, SAM, and SOM, outlining methodology and assumptions.
Agent
Layer-3 on-demand reference for the Layer-P memo phase. Defines the section template and downstream-consumer contract for the memo subagent output.
Agent
Lightweight PASS-cycle memo agent (v8.57.0+, Layer P). Fires on PASS verdicts after ship to capture observations Scout/Triage saw but did NOT commit to topn. Single-pass, $0.10–0.20 per cycle. Emits carryover-todos.json and memo.md — does NOT do retrospective/digest work (that's the FAIL/WARN retrospective agent's…
Agent
Merge-to-main promotion-readiness gate for the Evolve Loop (Evaluate archetype). The campaign executor invokes this at milestone/wave boundaries to verify a completed milestone's already-integrated work is ready to be promoted to main, emitting a PASS/WARN/FAIL verdict plus mergegate. signals. READ-ONLY — it never…
Agent
Metric tree evaluation agent for the Evolve Loop (Evaluate archetype). Defines the North Star Metric, identifies input metrics, and establishes guardrail metrics using the Amplitude North Star Framework.
Skill Claude CodeCodex
Use when designing or reviewing AI agent memory systems — covers memory type taxonomy (short-term, working, long-term), episodic vs semantic vs procedural memory, embedding-based retrieval with recency and importance weighting, memory compression and consolidation, cross-agent memory sharing, and anti-patterns like…
Skill Claude CodeCodex
Use when designing or reviewing multi-agent systems — covers orchestration topologies (hierarchical, flat, pipeline, hybrid), task decomposition, parallel fan-out/fan-in execution, result aggregation, shared context propagation, agent communication protocols, failure handling, token economics, and anti-patterns with…
Skill Claude CodeCodex
Use when designing or reviewing AI agents that must assess their own output quality — covers confidence scoring and calibration, chain-of-thought reflection, LLM-as-judge, hallucination self-detection, output quality verification, eval-driven development, and self-evaluation anti-patterns with examples in TypeScript…
Skill Claude CodeCodex
Use when designing multi-agent systems that must share consistent rules, values, and constraints across all agents — covers layered inheritance models, the CLAUDE.md pattern, override resolution, consistency enforcement, value propagation in parallel agents, conflict detection, and anti-patterns with TypeScript and…
Skill Claude CodeCodex
Use when designing or reviewing AI agent systems for cost and latency efficiency — covers context engineering (write/select/compress/isolate), prompt caching strategies, model routing by complexity, token budget management, context window techniques, prompt compression, and anti-patterns with TypeScript and Python…
Skill Claude CodeCodex
Use when reviewing LLM/AI-generated code — covers hallucinated APIs, plausible-but-wrong logic, authorization gaps, shallow error handling, over-abstraction, copy-paste context mismatch, missing edge cases, and outdated patterns with detection strategies and before/after examples in TypeScript, Python, and Go.
Skill Claude CodeCodex
Use when designing or reviewing AI/ML integration code — covers RAG pipeline design, prompt engineering, structured output with schema validation, tool use/function calling, LLM error handling, token budget management, hallucination mitigation, and AI/ML anti-patterns with examples in TypeScript and Python.
Skill Claude CodeCodex
Use when reviewing architecture or design for structural problems, when a codebase is hard to change or extend, when onboarding to a legacy system, or when recurring bugs suggest a deeper design flaw.
Skill Claude CodeCodex
Use when designing or reviewing rate limiting implementations — covers Token Bucket, Leaky Bucket, Sliding Window Counter, Fixed Window Counter, distributed Redis-based limiting with Lua scripts, response headers, client-side throttling/backoff, and anti-patterns across TypeScript, Go, Python, and Redis Lua.