AA dcl-semantic-drift-guard
Use this skill to detect semantic hallucinations and context drift in LLM outputs. Triggers when an agent or pipeline needs to verify that a generated response is faithfully grounded in a source document that was already provided inline — and has not fabricated, contradicted, or materially distorted any claims. Default mode (source_document provided directly) runs entirely inside the agent's own context with no network calls. Two clearly-labeled optional modes exist that do transmit data externally: kb_query (queries a remote RAG endpoint you configure) and an optional paid heuristic pre-check via Fronesis Labs' live DCL Trust Oracle MCP server. Do not use either optional mode with confidential, regulated, or sensitive source material without explicit confirmation from the user. Returns a tamper-evident DCL audit record with verdict IN_COMMIT or HALLUCINATION_DRIFT. Part of the DCL Skills verification suite by Fronesis Labs alongside DCL Policy Enforcer and DCL Sentinel Trace.
As a process A 84/100 · Runs to the end — weak spots: running it twice
How to improve
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 2. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
✓ No remarks against the Agent Skills spec
Process rating: all ten parameters 84/100
- 30Running it twice. 7 mutating operations with no state check
- 55Failures and branches. 1 branches
- 60Result and completion. Output format stated, no completion criterion
- 70Inputs and preconditions. Inputs and preconditions are listed
- 100Tools and files. No external tools needed
- 100Steps. 21 steps
- 100When it triggers. States when to use and when not to
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 2887 tokens
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low 12 top-level sections: this looks like several domains in one skill
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +3Description length 991: 120–800 characters recommended
- -218 emoji in the instructions: noise for the model
- +1No license
- +2Single-language instructions
- +4Description says when NOT to use the skill
- +4Structure: 24 headings
- +3Step-by-step instructions: 21 items
- +3Output format is stated explicitly
- +4Has examples (9 code blocks)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 86.