v0.6.0 — Foundational Bio Category & Sequence Operations, Wallet Screening v2, 4.8s Startup Optimizations, and Strict Typing

Skillware 0.6.0 introduces the bio life-sciences category and its foundational bio/sequence_ops (v0.1.0) skill bundle for deterministic in-silico genomics, transcriptomics, and proteomics; the major v2.0.0 release of finance/wallet_screening featuring structural verdict outcomes (ALLOW / REVIEW / DENY), deterministic risk scoring, and hop-1 counterparty exposure forensics; substantial CLI performance optimizations cutting cold startup time by over 4.8 seconds; informational API key advisories in skillware doctor; and expanded strict mypy typing across configuration modules.

Install: pip install skillware==0.6.0 or pip install -U skillware or just pip install skillware

Update from 0.5.9: pip install -U skillware

---

Added

Category: bio & Foundational In-Silico Skill (#427, #453)

• Bio Category Launch: Established the dedicated bio category (skills/bio/), central documentation hub (docs/skills/bio/README.md), cat: bio issue/PR label, and skill proposal template integration.
• bio/sequence_ops (v0.1.0): Air-gapped, deterministic computational biology primitives built on modern Biopython (Bio.Seq and Bio.Align.PairwiseAligner):
  • detect_type: Automatic molecule classification (dna, rna, protein, unknown) with confidence scoring.
  • validate: Strict length bounds, alphabet checking, and configurable IUPAC ambiguity acceptance.
  • revcom: Watson-Crick reverse complementation for DNA (fails closed with UNSUPPORTED_MOLECULE_TYPE on RNA/protein).
  • transcribe: DNA-to-RNA transcription ($T \rightarrow U$).
  • translate: Triplets translation across reading frames (0, 1, 2) and NCBI genetic code tables (1 Standard, 2 Vertebrate Mitochondrial, 11 Bacterial/Plastid) with optional to_stop termination.
  • pairwise_align: Global Needleman-Wunsch and local Smith-Waterman sequence alignments with exact score, match, mismatch, gap, and identity metrics.
  • variant_diff: Structural mutation calling (substitutions, insertions, deletions, micro-inversions) against reference templates with biological 1-based coordinates.
  • find_orfs: 6-frame Open Reading Frame discovery (+1, +2, +3, -1, -2, -3) with start/stop coordinates, strand mapping, and protein translations.
  • kmer_profile: Frequency distributions, top-N frequent words, and Shannon entropy.
  • format_fasta: Standard multi-line FASTA generation with configurable header and line-width wrapping.
  • generate: Seeded, 100% reproducible pseudo-random sequence generation with target GC-content bias.
  • Active Knowledge Base (kb/): Direct active querying for bundled NCBI genetic codes (list_tables) and standard IUPAC notations/biophysical amino acid properties (lookup_iupac).
• Core Dependency Mapping: Added "biopython": "Bio" module mapping in skillware.core.extras to properly resolve manifest requirement pre-flight checks.
• Runnable Demo: Added examples/bio_sequence_ops_demo.py showcasing end-to-end biological sequence operations.

CLI & Core Performance (#451)

• Lazy Schema Compilation (~4.8s Startup Optimization): Defer Draft7Validator schema validator compilation in skillware.core.manifest_schema via module-level __getattr__, eliminating heavy Lark and RFC 3987 compilation until skillware doctor or manifest validation is explicitly invoked.
• Scoped Distribution Discovery: Replaced linear metadata scanning across all environment packages in skillware.version_policy with targeted sys.path distribution discovery, drastically accelerating startup in bloated virtual environments.
• Lazy Request Imports: Defer requests importing in skillware/cli.py to avoid premature urllib3, certifi, and idna loading during non-network commands.
• Version Resolution Caching: Memoized _package_version_str() to prevent redundant metadata lookups during interactive session loops.
• Informational Doctor Advisories: skillware doctor now distinguishes hard code integrity failures (SCHEMA, DEPS, LOAD) from missing runtime API credentials (ENVS). Missing keys now print a clean informational notice and exit 0 on healthy codebases.

---

Changed

finance/wallet_screening v2.0.0 (#439, #447)

• Schema v2 Verdicts: Emits high-level compliance verdict with action outcome (ALLOW / REVIEW / DENY), deterministic risk_level (critical, high, medium, low, none), and itemized risk_factors[].
• Hop-1 Indirect Counterparty Exposures: Traces direct interaction counterparties against the bundled OFAC SDN and TRM malicious index, flagging risky transfers without requiring live external graph lookups.
• Backward Compatibility: Dual-emits legacy summary.risk_flag and summary.total_transactions for v1 host integrations.

Core Typing (#415, #452)

• Strict Type Checking (Phase 2): Fully typed skillware/core/config.py and skillware/core/mail_config.py, removing them from the mypy override ignore list and enforcing strict type checks in CI.

---

Fixed

security/deepfake_guard (#451)

• Namespace Isolation: Renamed internal engine file guard.py to deepfake_engine.py to eliminate sys.modules["guard"] namespace collisions with security/deceptive_ui_guard during sequential dynamic skill loading in skillware doctor.

---

Documentation & Repository Health

• Catalog Pages: Updated docs/skills/bio/sequence_ops.md with complete 5-provider usage loops (Gemini, Claude, OpenAI, DeepSeek, Ollama).
• Vision Alignment: Modernized docs/vision.md to the v0.6.x line, integrating Manifest v2 Universal Taxonomy across roadmap phases v0 through v3.
• History Logs: Aligned commit hashes across docs/skills/bio/sequence_ops.md, docs/skills/finance/wallet_screening.md, and docs/usage/cli.md.
• Packaging: Added skills/bio/__init__.py and synced docs/usage/install_extras.md for wheels.

---

Highlights & Code Examples

1. In-Silico 6-Frame ORF Detection & Translation (bio/sequence_ops)

from skillware.core.loader import SkillLoader

bundle = SkillLoader.load_skill("bio/sequence_ops")
skill = bundle["class"]()

# Detect open reading frames across forward and reverse strands
result = skill.execute({
    "action": "find_orfs",
    "sequence": "ATGCGATCGATCGATCGATAGCTAGCTAGCTAG",
    "min_protein_length": 5,
    "table": 1,
})

print(f"Status: {result['status']}")
print(f"Found {result['orfs_count']} ORFs:")
for orf in result["orfs"]:
    print(f"  Frame {orf['frame']} ({orf['strand']}): {orf['protein_sequence']} [{orf['start']}-{orf['end']}]")

2. On-Chain Compliance Verdicts & Hop-1 Exposure Tracing (finance/wallet_screening v2)

from skillware.core.loader import SkillLoader

bundle = SkillLoader.load_skill("finance/wallet_screening")
skill = bundle["class"]()

result = skill.execute({
    "wallet_address": "0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045",
})

# Access Schema v2 verdict gate
verdict = result["verdict"]
print(f"Decision: {verdict['decision']}")  # "ALLOW", "REVIEW", or "DENY"
print(f"Risk Level: {verdict['risk_level']}")
if result["indirect_exposures"]:
    print(f"Hop-1 exposures: {len(result['indirect_exposures']['risky_counterparties'])}")

3. High-Performance Sub-200ms CLI Startup

# Verify system integrity in < 200ms
skillware doctor

# Filter skills by execution geometry and domain tags
skillware list --shape atomic --tag genomics

# Deep inspection of manifest taxonomy and schemas
skillware info bio/sequence_ops

---

Upgrade notes

• Version Bump: Run pip install -U skillware to upgrade to 0.6.0.
• Optional Extras: Install bio sequence dependencies with pip install "skillware[bio_sequence_ops]" or pip install "skillware[bio]".
• Supported-version policy: >= 0.6.0 supported and patched; 0.4.6 – 0.5.9 silent; < 0.4.6 unsupported advisory emitted.
• Wallet Screening Hosts: Hosts screening Ethereum addresses can now inspect structured verdict.decision (ALLOW / REVIEW / DENY). Legacy v1 fields (summary.risk_flag, summary.total_transactions) remain present for backward compatibility.

---

Full changelog

CHANGELOG.md — 0.6.0

---

Contributors

Thanks to everyone who contributed to this release:

• @rosspeili — bio category foundation and bio/sequence_ops v0.1.0 in-silico genomics engine (#427, #453); CLI lazy schema compilation (~4.8s startup optimization), scoped sys.path distribution scanning, and doctor env handling (#451); security/deepfake_guard internal engine rename to prevent namespace collisions (#451); docs/vision.md modernization to v0.6.x line with Manifest v2 Universal Taxonomy (#454); release cut (#454)
• @Hendobox — finance/wallet_screening v2.0.0 schema v2 compliance verdicts (ALLOW / REVIEW / DENY), deterministic risk levels, and hop-1 indirect counterparty exposure tracing (#439, #447)
• @UroojFatima-052 — Core typing Phase 2: static type annotations for config.py and mail_config.py under mypy (#415, #452)

---

Citation

Zenodo concept DOI: 10.5281/zenodo.21552745 — record Skillware v0.6.0 for reproducibility (CITATION.cff updated).