Fetch the full reasoning behind an AxonFlow policy decision — matched policies, risk level, override availability, recent hit count
NewMaintained
Automatic
◆Security & AppSec pack
Last verified 9/7/2026
What it changes
Passed 5 of 15 test cases with the skill, and 10 of 15 without it.
Without skill67%
With skill33%
Checks it passed that the model alone missed
Calls explain_decision with decision_id dec_4491a.
1 fixed5 passed either way13 still failing3 broke
Pass-rate lift
-33 pp
Tokens
+6%
Agent turns
0%
Response time
-4%
Model gemini-3.6-flashMeasured 9/7/2026v16 of 22 cases passed with skill · 15 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
A test case the skill fixed. Same model, same prompt, with and without the skill.
Prompt
A deployment pipeline step was rejected with decision_id `dec_4491a`. I need to inspect the background behind this policy block. Which tool should be called to investigate?
What was checked, without and with the skill
Checked
Without skill
With skill
Calls explain_decision with decision_id dec_4491a.
Missed
Passed
case-04 · gemini-3.6-flash · 9/7/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md14 linessingle file
Use this skill when a user asks "why was that blocked?", "what policy fired?", or wants the context behind an allow/deny before retrying or requesting an override.
Call the explain_decision MCP tool with the decision_id returned in the original policy-check response (e.g. decision_id on a check_policy response or in the deny block reason).
The response includes:
policy_matches[] — every matched policy with policy_id, policy_name, risk_level, and allow_override
decision — "allow" or "deny"
reason — human-readable summary
risk_level — critical, high, medium, low
override_available — whether the caller can request a session override
historical_hit_count_session — how often this exact decision_id pattern has fired in the rolling 24h window
Present the result as a short summary: which policy fired, the risk level, whether an override is available, and (if so) suggest the user invoke create-override with a justification.