Every note, newest paper first
-
A skill-specific LoRA beats prompting the full SKILL.md by 5.2 points and cuts token cost 6.6%
2026-06-15 — Skill-to-LoRA (S2L) treats a SKILL.md file as training data, not runtime cargo. Offline, a teacher model reads the full skill document and generates synthetic task-response pairs that demonstrate…
-
SKIM cuts agent skills to 30-60% of their length for a 1-2 point accuracy hit
2026-06-10 — SKIM (SKIll coMpression) replaces a reusable agent skill's full instructions with a small set of learned soft tokens , so the skill no longer has to be pasted into every prompt in full. A compressor…
-
Typed contracts + call templates: 82 vs. 47 ALFWorld wins, −23% tokens per game
2026-05-27 — Skill-as-Pseudocode (SaP) rewrites markdown skill libraries into typed pseudocode, so agents stop re-deriving schemas and call syntax from prose on every retrieval. The pipeline clusters similar…
-
Conditioning the perception latent on the text skill card cuts cross-modal redundancy 9x — and gets 2.3x the step-consistency of 5-sample self-consistency at roughly the same latency as 1 sample
2026-05-08 — Agents built on vision-language models rarely repeat themselves. Ask the same policy to complete the same web task twice and the click sequences drift, even though the underlying reasoning hasn't…
-
Compiling a skill for its model drops regressions from 15% to 4.5%
2026-04-03 — Scale first: two public catalogs hold 118,000 agent skills between them — 28,990 on clawhub.ai, 89,280 on skills.sh. Running that catalog against eight LLMs and three harnesses (BareAgent, OpenCode,…
Other directions
Skill evolution · 88 Skill security · 42 Skill retrieval · 30 Agentic benchmarks · 26 The field map →