Pith Number

pith:PNWKDRK6

pith:2026:PNWKDRK6TJADZZFDVLT4TTKRTK

not attested not anchored not stored refs resolved

SKILL0: In-Context Agentic Reinforcement Learning for Skill Internalization

Chengcheng Han, Jinyang Wu, Jun Xiao, Qi Gu, Weiming Lu, Xunliang Cai, Yongliang Shen, Yueting Zhuang, Zhengxi Lu, Zhiyuan Yao

A curriculum of progressively withdrawing skill context during reinforcement learning lets agents internalize procedural knowledge into their parameters for zero-shot task completion.

arxiv:2604.02268 v2 · 2026-04-02 · cs.LG

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{PNWKDRK6TJADZZFDVLT4TTKRTK}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

SKILL0 achieves substantial improvements over the standard RL baseline (+9.7% for ALFWorld, +6.6% for Search-QA, and +10.1% for WebShop), while maintaining a highly efficient context of fewer than 0.5k tokens per step.

C2weakest assumption

The Dynamic Curriculum can accurately identify on-policy helpfulness of individual skill files and that progressive context withdrawal produces genuine internalization rather than superficial adaptation to the training distribution.

C3one line summary

SKILL0 uses in-context RL with a dynamic curriculum to internalize skills into LLM parameters, yielding performance gains on agent benchmarks with under 0.5k tokens per step.

References

3 extracted · 3 resolved · 0 Pith anchors

[1] If any required knowledge is missing or uncertain, youMUSTcall a search engine to get more external information using format:<search> your query </search>

[2] Additionally, select an image compression factor larger than 1.0 for the next image

[3] 2.<search>...</search>or<answer>...</answer> 3.<compression>...</compression> Figure 12: Prompt template used by SKILL0 for the Search-based QA task environment

Formal links

2 machine-checked theorem links

Cited by

15 papers in Pith

Terminal-World: Scaling Terminal-Agent Environments via Agent Skills

A Comprehensive Survey on Agent Skills: Taxonomy, Techniques, and Applications

Dynamic Skill Lifecycle Management for Agentic Reinforcement Learning

SkillsVote: Lifecycle Governance of Agent Skills from Collection, Recommendation to Evolution

Test-Time Learning with an Evolving Library

Receipt and verification

First computed	2026-05-20T00:00:37.543511Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

7b6ca1c55e9a403ce4a3aae7c9cd519a9939cc4b3128b91dbb3a1d31dacb9494

Aliases

arxiv: 2604.02268 · arxiv_version: 2604.02268v2 · doi: 10.48550/arxiv.2604.02268 · pith_short_12: PNWKDRK6TJAD · pith_short_16: PNWKDRK6TJADZZFD · pith_short_8: PNWKDRK6

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/PNWKDRK6TJADZZFDVLT4TTKRTK \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 7b6ca1c55e9a403ce4a3aae7c9cd519a9939cc4b3128b91dbb3a1d31dacb9494

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "04a57bf11d3aeebb0b3b70c216caa9dc750cd88ef1914062cde1ca5672ded972",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2026-04-02T17:03:05Z",
    "title_canon_sha256": "7241591e3f1c7b6a521e82ecccc8cae7ff21579425c46f73fc040b05d46907ae"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2604.02268",
    "kind": "arxiv",
    "version": 2
  }
}