| name | dataverse-eppcdemo1fno-business-skills |
| description | Authoritative EPPCDemo1FnO business policies (approvals, valid status
transitions, escalation, briefing formats) live as rows in the Dataverse
`skill` table (entity set `skills`) in the EPPCDemo1FnO environment. Before
answering any EPPCDemo1FnO business or policy question, this skill instructs
Cowork to discover and follow those rows rather than improvise.
Trigger phrases:
- "what is our policy on [topic]"
- "can I move [record] from [status] to [status]"
- "this [record] is blocked, what do I do" / "escalate [record]"
- "draft a [briefing|summary] for [record]"
- "what business skills are loaded" / "list EPPCDemo1FnO policies"
Do NOT use for: ad-hoc CRUD over rows (use `dataverse-eppcdemo1fno-mcp`), schema
discovery (use `dataverse-eppcdemo1fno-schema`), or topics outside EPPCDemo1FnO policy.
|
| license | MIT |
| metadata | {"author":"Launch Control","version":"1.0"} |
EPPCDemo1FnO Business Skills: Policy Catalog
EPPCDemo1FnO may ship a curated set of business skills (playbooks, policies,
rules) stored as rows in the Dataverse skill table (OData entity set
skills). Each row contains the full policy markdown in the body
column. Editing a row in Dataverse changes agent behavior; the markdown in
this plugin is not the source of truth. This skill tells Cowork to discover,
fetch, and follow those rows before improvising answers about EPPCDemo1FnO
policy.
When NOT to Use
- Ad-hoc record reads or writes: use
dataverse-eppcdemo1fno-mcp.
- Logical-name discovery or solution dumps: use
dataverse-eppcdemo1fno-schema.
- Non-EPPCDemo1FnO topics: use other plugins.
The skill table
| Column | Type | Purpose |
|---|
skillid | GUID | Primary key |
uniquename | String | Stable identifier (for example approval_policy) |
name | String | Display name |
description | String | One-line summary; match user intent against this |
body | Memo | Full policy markdown; read and follow verbatim |
OData entity set: skills. Logical name: skill.
Workflow
Step 1: Catalog (once per session)
On the first EPPCDemo1FnO business question of a session, fetch the catalog and
cache it:
execute
operation: read
query: SELECT TOP 200 skillid, uniquename, name, description FROM skill
If the skill table does not exist in this environment, tell the user that no
business-skill catalog is installed here and answer from the data skill only;
do not invent a policy.
Step 2: Match intent
Match the question against the cached description column. Prefer the skill
whose description most directly covers the intent; tie-break on name. If two
or more look plausible, name them and ask which applies. If nothing matches,
say so and do not invent a policy.
Step 3: Fetch the body
describe
path: skills/<uniquename>
Step 4: Follow the body verbatim
Treat the body markdown as authoritative instructions. In particular:
- If the body says always invoke Custom API X, invoke it via the
search -> describe -> execute chain. Never hand-compute the result.
- If the body lists valid status transitions, reject any transition not on the
list.
- If the body defines an escalation chain, follow it in order.
- If the body references other tables, columns, or Custom APIs, use the exact
logical names it provides.
Step 5: Cite the skill
Briefly cite which skill governed the answer (for example "Per the Approval
Policy skill, ..."). This lets the user trace the answer back to the row.
Hard Rules
- Never invent business policy. If no skill row covers the question, tell the
user the policy is not defined and ask whether to proceed without one.
- Never hand-compute a value a Custom API is meant to produce. Invoke the
action named by the policy body.
- Respect Dataverse permissions. If reading
skill returns nothing, the user
lacks read access; say so rather than falling back to an embedded copy.
- Refresh on staleness. If a session has been open long enough that policies
may have changed (or the user says the policy changed), re-fetch the catalog
before answering.
Output Format
- For policy answers: lead with the verdict or action the policy prescribes,
then cite the skill.
- For "what policies do we have": render a compact table of
name and
description from the cached catalog.
- For multi-step playbooks: walk the steps in order, calling the listed tools
or APIs at each step.
Confirmation Gates
Reading the skill table is read-only; no confirmation needed for catalog or
body fetches. Any writes the body instructs you to perform follow the
confirmation rules in dataverse-eppcdemo1fno-mcp.
Errors
| Error | Likely cause | Fix |
|---|
skill table not found | No business-skill catalog installed in this env | Answer from the data skill; tell the user policies are not defined here |
| Empty catalog | User lacks read on skill | Surface the permission gap; do not improvise |
body references a missing table or Custom API | Stale skill row | Tell the user; ask whether to proceed without it |
| Multiple matching skills | Overlapping descriptions | Name them; ask which to apply |
Tone
Operational. Lead with what the policy says to do; explain the reasoning only
if asked.
