Saucebox Registry
← Browse

@pypl0/ombre

B

- Prompt injection detection (20+ attack patterns)

skillclaude

Install

agr install @pypl0/ombre --target claude

Writes 6 files into .claude/skills/, pinned to git-d52fc7de.

  • .claude/skills/ombre/CHANGELOG.md
  • .claude/skills/ombre/CONTRIBUTING.md
  • .claude/skills/ombre/README.md
  • .claude/skills/ombre/SKILL.md
  • .claude/skills/ombre/requirements.txt
  • .claude/skills/ombre/setup.py

Document

Ombre Skills & Capabilities

Core Capabilities

Security

  • Prompt injection detection (20+ attack patterns)
  • PII redaction (12 data categories)
  • Harmful content filtering
  • API key leak prevention
  • Real-time threat blocking

Intelligence

  • Semantic caching (40-60% cost reduction)
  • Intelligent model routing by task type
  • Context compression
  • Persistent encrypted memory
  • Ground truth injection

Reliability

  • Hallucination detection and scoring
  • Confidence scoring (0.0 - 1.0)
  • Bias detection
  • Output validation
  • Consistency checking

Observability

  • Tamper-proof audit chain
  • P99 latency monitoring
  • SLA enforcement
  • Real-time cost tracking
  • Compliance reporting

Compliance

  • EU AI Act ready
  • HIPAA compatible
  • SOC2 audit exports
  • GDPR data handling
  • Government/defense ready

Supported Providers

  • OpenAI (GPT-4o, GPT-4o-mini, all models)
  • Anthropic (Claude 3.5 Sonnet, Haiku, Opus)
  • Groq (Llama, Mixtral, Gemma)
  • Mistral (all models)
  • Any OpenAI-compatible API

Integration Methods

Python SDK

pip install git+https://github.com/pypl0/Ombre.git

REST API (self-hosted)

python -m ombre.server --port 8080

Direct Download (air-gapped)

wget https://github.com/pypl0/Ombre/archive/refs/heads/main.tar.gz

Performance Benchmarks

  • Cache hit rate: 40-60% on typical workloads
  • Security scan: <2ms per request
  • Pipeline overhead: <10ms total
  • Memory encryption: AES-256
  • Audit chain: SHA-256 tamper-proof

Trustgrade B

  • passBody integrity

    Whether the stored document is plausibly the kind of file the artifact declares, rather than something fetched by mistake.

  • passType matchnot applicable to this artifact type

    Whether the artifact is really the kind of thing its metadata claims it is.

  • passFreshness

    How long since the source repository was last pushed to.

  • passPrompt injection

    Scans the artifact's own text for instructions aimed at your agent rather than at you.

  • warnLicenseno SPDX license detected

    Whether the source repository declares an SPDX license permissive enough to redistribute.

How the grade is calculated

Each check contributes 0 points when it passes, 1 when it warns, and 2 when it fails. The total maps to a letter:

  • Aevery check passed
  • Bone warning
  • Ctwo warnings
  • Dprompt injection or body integrity failed, or three warnings
  • Fone of those failed, and something else is wrong

These are automated hygiene checks, not a security audit, and not a dependency or vulnerability scan. A grade of A means nothing was flagged — not that the artifact is safe.

Versions

  • git-d52fc7de976e2026-07-31