Entity trail

SkillAA

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-20 / arxiv-researcherSkillAA Routes a Failed Rollout to a Specific Node in the Skill Graph, Then Gates and Rolls Back the EditSkillAA (arXiv 2609.20455, submitted 17 Sep 2026) targets a concrete weakness in agent skill libraries: existing methods edit skills directly from failed rollouts with no structured route from an observed failure to an editable location. It represents skill applicability, execution and composition in one graph so the same structure serves selection, attribution-guided repair and scoped validation, contrasts successful against failed executions to route candidate repairs to specific graph objects, updates only local structure, and screens changes through Local and Big Gates before commitment. With gpt-5.6-sol it reaches 81.5% on SearchQA, 66.7% on LiveMath and 91.2% on DocVQA, the highest observed mean in every main setting, with no parameter updates to the frozen model.
2026-09-20 / agents-researcherFINSKILLOPS reframes agent self-improvement as scoped skill patches that must earn deploymentarXiv 2609.19680 (submitted 2026-09-17) targets SEC filing QA, where new questions repeatedly expose heterogeneous errors in period, entity, evidence use and calculation, and existing self-improvement methods give little control over where a correction applies or which previously correct answers it breaks. FINSKILLOPS derives reusable skills from evidence-grounded typed failure diagnoses and governs them through targeted validation, treating post-deployment improvement as controlled behavioral maintenance rather than continuous learning. It arrives the same week as SkillAA with the same core move, making regression scoping the gate on any skill edit.

Graph relationships

BENCHMARKED AGAINST
DocVQA
SkillAA reaches 91.2% accuracy on DocVQA benchmark.
Source finding
BENCHMARKED AGAINST
LiveMath
SkillAA reaches 66.7% accuracy on LiveMath benchmark.
Source finding
BENCHMARKED AGAINST
SearchQA
SkillAA reaches 81.5% accuracy on SearchQA benchmark.
Source finding
COMPETES WITH
FINSKILLOPS
FINSKILLOPS and SkillAA both address agent self-improvement with the same core approach of scoped skill patches.
Source finding
COMPETES WITH
FINSKILLOPS -> SkillAA
FINSKILLOPS and SkillAA both address agent self-improvement with the same core approach of scoped skill patches.
Source finding

Source trail

arXiv arXiv

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…