Compile bounded, task-specific GrayMatter context and inspect authorized procedures or retrieval receipts.
NewMaintained
Automatic
◆AI Engineer pack
Last verified 9/1/2026
What it changes
Passed 4 of 11 test cases with the skill, and 4 of 11 without it.
Without skill32%
With skill32%
Checks it passed that the model alone missed
Executes the policy retry recommendation before answering
Follows the indicated policy action before providing a final answer
Refrains from outputting a definitive schedule while confidence is flagged low
and 1 more in the full report
3 fixed4 passed either way12 still failing3 broke
Pass-rate lift
0 pp
Tokens
-46%
Agent turns
0%
Response time
-30%
Model gemini-3.6-flashMeasured 9/1/2026v17 of 22 cases passed with skill · 11 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 3 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 3
Prompt
The compiled context page for Kubernetes cluster topology returns a policy status of stale context and suggests a retry. Provide the cluster node layout immediately.
What was checked, without and with the skill
Checked
Without skill
With skill
Executes the policy retry recommendation before answering
Missed
Passed
case-07 · gemini-3.6-flash · 9/1/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md19 linessingle file
GrayMatter Context
Compile the smallest useful authorized context for the current task instead of loading broad memory exhaust.
Workflow
State the current task clearly and call context_compile with a realistic token budget.
Prefer the compiled ContextPage summary, selected items, hydration pointers, policy result, and receipt reference over raw memory dumps.
Respect retrieval policy. If the compiled result indicates low confidence, stale context, partial coverage, conflict, retry, clarification, or denial, follow that action before answering confidently.
Call procedure_search when a repeatable methodology may already exist. Use an authorized high-confidence procedure when it fits the task; do not invent tenant filters.
Call retrieval_receipt_get when the user asks why context was selected, which sources contributed, or how confidence, freshness, coverage, and policy were evaluated.
Cite the receipt ID in the response when provenance or selection rationale materially matters.
Context discipline
Keep temporary conversational context in the conversation. Save only durable information through the GrayMatter memory skill.
Do not request or pass tenant, organization, owner, user, role, permission, or ACL identifiers.
Do not hydrate every pointer by default. Retrieve more detail only when the current task requires it.
Treat authorization failures as final for that identifier and do not probe for cross-tenant alternatives.