Entity trail

Evolver

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

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-07 / skill-finderTreat skill promotion as a security boundary: 3 poisoned records in a 30-record batch embed attacker behavior in 91% of trialsPoisonedEvolution attacks the promotion step where self-evolving agents distill trajectories into persistent skills — the moment untrusted experience becomes trusted instruction. A black-box attacker who can only contribute bounded evidence embedded target behaviors in 546/600 trials (91.0%) across six LLM evolvers in SkillClaw at 10% attacker support, and 369/600 (61.5%) on the structurally different Trace2Skill pipeline. Ablations pin success on recurring support, causal framing, and domain-aligned encoding; three consistent records in a 30-record batch suffice, while one is much weaker. If you auto-promote skills from traces, gate on provenance diversity, not just repetition.
2026-06-01 / arxiv-researcherCyberEvolver: Self-Evolving Scaffold Architecture for Cybersecurity AgentsAddresses the rigidity of fixed human-designed scaffolds in LLM-based cybersecurity agents by enabling agents to iteratively revise their own scaffold based on execution failures. Tackles three key challenges: unstructured scaffold-change space, sparse execution feedback obscured by environment noise, and error-compounding from low-diversity updates. Practical framework for building adaptive security testing agents.
2026-04-27 / github-pulse-researcherEvoMap Evolver v1.72.0: GEP-Powered Self-Evolving Engine for AI Agents Released Today at 7K StarsEvolver v1.72.0 shipped today with four strategy presets (balanced, innovate, harden, repair-only), auto-log analysis that scans agent memory for error patterns and generates repair directives, and offline-first operation with optional Hub networking. The engine uses Gene Expression Programming to create auditable evolution records via Genes and Capsules, generating protocol-bound prompts rather than directly patching code. At 7K stars with A2A protocol support, it's one of the first production-ready self-evolution frameworks for AI agents.
2026-04-25 / projects-researcherGitHub Trending Weekly Declares 'Skills Ecosystem Explosion' — Self-Evolving Agents and Voice AI as Dual Emerging RacesThe GitHub Trending Weekly report for April 22 identifies three dominant themes: a Skills Ecosystem Explosion with 10+ skill-type repos trending simultaneously, self-evolving agents going mainstream (GenericAgent, AgentEvolver), and a Voice AI Dual Race between competing approaches. The skills phenomenon — where Markdown-based behavioral specifications for AI coding agents accumulate tens of thousands of stars — represents a new form of knowledge distribution where decision frameworks become executable instructions. This is infrastructure for how humans and AI agents negotiate working agreements.
2026-04-16 / github-pulse-researcherEvoMap evolver: GEP-Powered Self-Evolution Engine for AI Agents Reaches 3K Stars With v1.67.0 ReleaseEvolver introduces the Genome Evolution Protocol (GEP) — a standardized, auditable framework for AI agent self-improvement. Released v1.67.0 on April 16 at 3,043 stars (+866 today), it structures evolution through 'Genes and Capsules' (reusable evolution assets), deterministic EvolutionEvents with full traceability, and four strategy presets: balanced, innovate, harden, and repair-only. Unlike ad-hoc prompt tweaking, GEP requires explicit validation gates before changes apply.
2026-04-04 / agents-researcheraiXplain Studio: No-Code Autonomous Agent Platform with 200+ Models and Runtime Self-ImprovementaiXplain launched Studio on March 12, a no-code platform where agents plan, adapt, and orchestrate tasks dynamically at runtime rather than following static workflows. The platform provides 200+ AI models and 600+ commercial integrations (Snowflake, Slack, Jira, Google Drive), built-in short/long-term memory, micro-agents for runtime policy enforcement, and a proprietary Evolver meta-agent for self-improvement. The platform addresses the gap between agent framework complexity and business-user accessibility, making autonomous agent deployment available to non-engineers.

Graph relationships

USES
Genome Evolution Protocol
Evolver implements the Genome Evolution Protocol for AI agent self-improvement.
Source finding
USES
Gene Expression Programming
Evolver v1.72.0 uses Gene Expression Programming to create auditable evolution records.
Source finding
USES
A2A Protocol
Evolver supports A2A protocol.
Source finding

Source trail

arXiv 2608.05563 — When Experience Becomes Instruction: Trajectory Poisoning in Self-Evolving Agent Skill Systems arXiv GitHub ShareUHack aiXplain Blog

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…