All authors
Claude Skills by ArttuAn
github.com/ArttuAn18 skills0 installs18 views
- Docs<One sentence, under 220 characters. What the IDE will build, and the thing that makes it hard. Start with a verb: Build a ... harness from a spec — ...>Votes: 0GitHub stars: 2
- Code AgentBuild a sandboxed coding-agent harness from a spec — workspace confinement, bounded command execution, reviewable diffs and an undo pathVotes: 0GitHub stars: 2
- Multi AgentBuild a multi-agent orchestrator harness from a spec — role workers with isolated context, structured handoffs, bounded fan-out and honest cost accountingVotes: 0GitHub stars: 2
- Plan And ExecuteBuild a plan-and-execute agent harness from a spec — planner, executor and verifier with bounded replanning and a plan you can inspectVotes: 0GitHub stars: 2
- Rag MemoryBuild a RAG / long-term-memory agent harness from a spec — chunking, embeddings, hybrid retrieval and cited answers that stay groundedVotes: 0GitHub stars: 2
- ReactBuild a ReAct (reason-act-observe) agent harness from a spec — a tool-calling loop with the guards that keep it terminating, testable, and recoverableVotes: 0GitHub stars: 2
- ReflexionBuild a Reflexion agent harness from a spec — act, evaluate, reflect, retry, with episodic memory of past mistakes and an evaluator that can actually failVotes: 0GitHub stars: 2
- Self EvolvingBuild a self-evolving agent harness from a spec — run, evaluate, reflect, apply, gated by a test suite the agent cannot edit and a rollback it cannot skipVotes: 0GitHub stars: 2
- ConsolidationBuild the background pass that keeps memory healthy: dedupe, merge, promote episodes to facts, decay and prune - with lineage and a dry runVotes: 0GitHub stars: 2
- EpisodicBuild an episodic memory: an append-only log of what the agent did, with outcomes, recalled by similarity and recencyVotes: 0GitHub stars: 2
- EvalBuild the probe harness that proves a memory system works: recall@k, staleness, leak gates, negative controls, and longitudinal driftVotes: 0GitHub stars: 2
- GraphBuild a knowledge-graph memory: entity extraction, conservative entity resolution, edge storage, and seed-and-traverse retrievalVotes: 0GitHub stars: 2
- ProceduralBuild procedural memory: playbooks the agent learns from repeated episodes, matched by precondition, replayed and demoted on failureVotes: 0GitHub stars: 2
- ScratchpadBuild file-backed working memory: a structured notes file the agent reads and rewrites each turn, atomic, diffable, and editable by a humanVotes: 0GitHub stars: 2
- SemanticBuild a semantic fact store: extraction with a gate, entity keying, contradiction resolution, confidence and provenanceVotes: 0GitHub stars: 2
- SharedBuild a multi-agent blackboard: namespaced writes, work leases, compare-and-set conflict detection, and bounded visibility between workersVotes: 0GitHub stars: 2
- TemporalBuild a bitemporal memory: facts with validity intervals and belief time, point-in-time queries, invalidation instead of deletionVotes: 0GitHub stars: 2
- WorkingBuild the agent's short-term memory: token budgeting, compaction, eviction policy, and pinned turns that survive itVotes: 0GitHub stars: 2