Decision-Grade Reasoning (DGR)
v1.0.4Audit-ready decision artifacts for LLM outputs — assumptions, risks, recommendation, and review gating (schema-valid JSON).
⭐ 3· 3k·7 current·7 all-time
byKhazretgali Sapenov@sapenov
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (produce auditable, schema-valid decision artifacts) aligns with the provided files (SKILL.md, prompt.md, schema.json, examples). No binaries, env vars, or installs are requested — all consistent with an instruction-only formatting/templating skill.
Instruction Scope
Runtime instructions are narrowly scoped to producing a JSON artifact that conforms to schema.json, asking for clarifications when inputs are missing, and to avoid fabricating sources or chain-of-thought. It also directs generating UUIDs and computing a stable query hash (e.g., sha256 of the user query) — operations that are reasonable but will surface whatever user input is provided. Note: the skill will encode decision content into artifacts, so sensitive inputs become part of that artifact.
Install Mechanism
No install spec or code artifacts that execute on disk; instruction-only skill — lowest-risk install profile.
Credentials
Requires no environment variables, credentials, or config paths. The lack of requested secrets is proportionate to the described functionality.
Persistence & Privilege
always:false and normal model invocation behavior. The skill does not request persistent system presence or modify other skills. Users should, however, consider how/where generated artifacts are stored because they may contain sensitive decision data.
Assessment
This skill is instruction-only and internally consistent: it only formats reasoning into a strict JSON schema and requests no credentials or installs. Before using it, consider: (1) any sensitive or personally identifiable information you include in the question will appear verbatim in the artifact — ensure secure storage and retention policies; (2) for legal/medical/financial or safety-critical decisions rely on human experts (the skill explicitly gates review_required for high-stakes cases); (3) validate produced artifacts against schema.json in your pipeline and spot-check outputs for hallucinated facts or missing clarifications; and (4) test the skill with representative inputs to confirm the agent implements the Like a lobster shell, security has layers — review code before you run it.
latestvk97fsndwn3ynf1958rcqxcdqm180gp6h
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🧭 Clawdis