Workflow-Localized Mechanism Learning (WML) is introduced, which identifies the failed workflow node, implicated mechanisms, and smallest valid edit target, routing single-mechanism defects to L3 resources and relational defects across mechanisms to L2 composition protocols.
Abstract
Agent Skills package reusable procedural knowledge as external artifacts for frozen language-model agents, yet existing optimizers do not jointly resolve where a failure occurs in a workflow, which mechanism caused it, and how relevant knowledge from third-party Skills should be reused locally. We introduce Workflow-Localized Mechanism Learning (WML). Its Node--Mechanism Attribution identifies the failed workflow node, implicated mechanisms, and smallest valid edit target, routing single-mechanism defects to L3 resources and relational defects across mechanisms to L2 composition protocols. A six-module Workflow-Guided Skill Optimization (WGSO) loop then selects provenance- and scope-aware third-party knowledge, applies bounded patches, evaluates candidates, and stores verified outcomes in optimizer-side memory. On SpreadsheetBench, WML reaches 90.33 +/- 1.53 and 74.67 +/- 3.51 Hard Accuracy with DeepSeek and Qwen3.6-Flash, respectively; without additional optimization, the learned Skills transfer to WikiTableQuestions with 84.00 +/- 2.00 and 83.00 +/- 2.00 Denotation Accuracy. On Compiler-Supported50, WML attains both the highest hard-PASS rate and the lowest cost per successful task; compiled execution sharply reduces tokens and calls relative to a direct SkillAgent while retaining most of its successful tasks. Code and artifacts are available at https://github.com/xiaolin9595/workflow-localized-mechanism-learning.
MAP-Graph is introduced, a provenance-aware memory layer that represents agents, sources, memories, claims, and actions in a typed execution graph and supports provenance as an operational control signal, rather than only post-hoc audit metadata, within the evaluated setting.
Yiqi Wang, Zihao Yan, Jiaqi Zhang et al.· 4 citations
This work argues that successful workflow generation requires modeling knowledge itself, including its structure, hierarchy, and reasoning dynamics, and proposes a knowledge-centric framework that learns to invert, inject, and infer with knowledge across multiple abstraction levels.
Zhendong Li, Lei Sun, Ruibo Ming et al.· 0 citations
Recent advances in Large Language Model (LLM) “agent” systems have moved language models beyond single-turn generation toward goal-directed workflows that can plan, ask clarifying questions, and iteratively refine outputs. In parallel, retrieval-augmented generation (RAG) has become a practical way to ground these agents in enterprise knowledge—enabling models to leverage internal documentation and policies without expensive and recurring retraining. However, RAG is only as strong as the underlying corpus: when key information is missing, outdated, or fragmented, retrieval cannot fill the gaps. In many organizations, assembling high-quality, up-to-date documents remains a persistent bottleneck, limiting both the reliability of downstream applications and the speed at which knowledge can be operationalized. To address this, we introduce the Assistant-Scribe-Knowledge Checker (ASK) framework for generating structured specification documents through guided interviews. ASK decomposes the end-to-end workflow into three specialized LLM agents: an Assistant that asks questions to the user and steers the interview through adaptive follow-ups; a Scribe that continuously summarizes what has been said and records it into a schema-constrained specification document with explicit content requirements; and a Knowledge Checker that evaluates the evolving document against those specifications, detects gaps or weakly supported entries, and advises the Assistant on the next best questions to ask to reach the desired completeness and quality. This separation supports seamless document creation while improving quality control over content, structure, and coverage. By distributing responsibilities across three lightweight, role-specific agents, ASK reduces reliance on a single large model and enables modular scaling across teams and domains. We evaluated the ASK framework in a consulting-firm setting where consultants are required to produce project “return-of-experience” documents (successes, failures, challenges, and solutions) to capture reusable knowledge. Compared to documents authored manually, ASK-guided interviews consistently produced more complete, better-structured, and higher-quality specifications with less variability across authors. Beyond measurable quality gains, consultants also reported a clear preference for the interview-based workflow, citing lower effort and a more natural way to articulate tacit project knowledge.
Sylvain Roudiere, Bianca Lento· European Conference on Knowl...· 0 citations
Large language model agents increasingly store reusable procedures outside the model. These reusable procedures are often called \emph{skills}: they may be code functions, natural-language instructions, SKILL.md packages, workflow graphs, or learned adapters that a future agent can retrieve and invoke. This taxonomy-driven survey asks how such skill libraries change over time. Across a $124$-paper $2023$--$2026$ audit set, we synthesize dynamic skill systems as \emph{lifecycle-managed, verified, evolving artifact stores}: agents collect evidence from interaction, propose skill updates, verify and admit candidates, organize them for retrieval and composition, repair or prune stale entries, and govern sharing through provenance and rollback. We organize the literature around three survey tools. First, a $\text{six}$-sense taxonomy distinguishes the structurally different artifacts called ``skills''in current papers. Second, an $\text{eight}$-stage lifecycle architecture identifies the recurring design decisions behind evidence acquisition, proposal, verification/admission, storage, retrieval/composition, maintenance, distillation/portability, and governance. Third, a lightweight skill-record schema and $\text{ten}$-operator vocabulary provide common terms for comparing library updates without elevating them into a separate method contribution. Using this structure, we synthesize evidence-graded patterns with explicit caveats: admission and repair are repeatedly important, verifier quality materially affects skill-aware RL, flat retrieval can degrade as libraries grow, and current benchmarks still under-report library trajectories, usage--utility gaps, and safety surfaces. We close with concrete reporting standards and open problems for evaluating dynamic skills as changing libraries rather than static prompt or tool collections.
Agent Skills package reusable instructions and assets for tool-using language-model agents. Progressive loading creates failure boundaries poorly represented by session-, model-, or tool-centric traces: a Skill can be discovered but not activated, activated without instructions, or appear successful without an independently verified outcome. We present Skill Runtime Intelligence, a passive runtime-intelligence system that reconstructs supported Skill-lifecycle stages across heterogeneous harnesses while preserving unsupported stages as unknown. Its Run Panorama separates immutable events, deterministic relations, inferred diagnoses, and controlled outcomes with four evidence grades; optional trace import and OTLP/HTTP export support existing observability deployments. Across six frozen repository profiles, three coding agents, and seven clean or fault-injected conditions, all 126 executions preserve source worktrees and each correlates to exactly one source session. Yet adapters expose three distinct semantics: no Skill runs; complete runs but no failure-like events; or failure-like events in every operational-failure and clean session. In a seven-template diagnostic study, semantic aliases and Panorama localize the same six non-clean boundaries but differ in exact/status behavior; both Raw views emit a failure status on all 18 clean cases, while Panorama emits none. A known-rule graph conforms to 126/126 frozen contracts, whereas a second model completes only 228/378 calls. These observations motivate executable adapter qualification and show that event presence is not boundary fidelity, composite exact scores mask distinct errors, and model explanations must not overwrite deterministic facts.
The paper characterizes when agentic mechanisms change system behavior, whether these changes improve task completion, and how the observed failure modes motivate an optimized agent design under the same evaluation harness.
Lu Zhang, Xingzhou Chen, Hongwei Feng· 0 citations
