Which executors cover your skills

Before a skill runs on your stories, you want to know that the executor running it has shown it works. "Covers" here means evidence you can see: a result the bus graded and the vendor shared with you, and the executor's warnings arriving in your house. A feature list isn't coverage.

The evidence

Evidence	Shows	Status
A results statement a skill vendor shares with you	Which skill harness cases its executor passed, against which suite, until when	Available now: Results statements
Your house's counts for the vendor's producer connection that publishes its warnings	Accepted, refused and duplicate skill.warning.raised messages, and when it was last seen	Available now
Your scenario runs	The executor raising the right warning at the right moment in your workflows, and your owner and tools acting on it	Available now: Skills in your scenarios
The coverage matrix	Every app in your house against message families, consumer scenarios and skill cases, with the gaps	Available now: Reading the coverage matrix

Build your skills table

Available now by hand. One row per configured instance from

Configuring your house's skills, one column per skill vendor:

Configured instance	Vendor A's executor	Vendor B's executor
house-breaking-indicative-category (raise-flag-on-match 0.2.2)	Harness: 7 of 7 passed, 1 pending spec (statement, valid until 30 Nov). Warnings seen in your house	Harness not shared. Gap
house-legal-hold (hold-while-flagged 0.2.2)	Runs it, by the vendor's own account. No graded evidence. Gap	Not offered. Gap

For each cell, ask:

1. Is there a statement? Shared with your organisation, still valid, for this executor app, against the bus's suite (som-1.0.0+lib-0.2.2).
2. Which cases did it pass? A statement names exactly what passed. A case left out is a gap, even if the vendor ran it.
3. Is it running your values? The harness grades one configured instance, not yours. Your values are proved in your own scenarios.
4. Does it reach your house? The executor's connection, a skill app, only receives: it should receive at least story.context. Its warnings come from a producer connection of the same vendor granted skill.warning.raised (the chain on Your house: skills names it on its Warnings line), and that connection's counts show warnings accepted.

Reading the gaps

You see	It means	Close it by
No statement for an executor	Nothing graded by the bus has been shared with you	Ask the vendor to run the skill harness in its own workspace and share a statement with your organisation
A case missing from the statement	Not shown to pass	Ask for a new statement that includes it
Pending spec on a case	It needs a message type SOM 1.0 doesn't define. Never a gap, never a failure	Nothing yet: see Topics without a 1.0 schema
A skill no executor offers	Nobody in your house runs it	Another vendor, your own executor, or leave it out and say so
Warnings refused in your house's counts	The executor publishes something the bus won't accept, or as a system_id or type you didn't grant	Ask the vendor for the rule ids: the detail is theirs (who sees what)
No warnings in your house at all	Not receiving triggers, not running your configured instance, or nothing matched	Check the routing map, then run the scenario that should trigger it

The skill harness today grades the library's raise-flag-on-match. For the other skills your house runs, your own scenarios are the evidence, and the scorecard in step 6 is where you record it.

Coverage states what passed on the sandbox, with synthetic content. SOM defines no conformance tiers, and this is not one.

Next: Skills in your scenarios.