adobe/workfront
Adobe Workfront skills for AI agents. Includes Workfront Planning (WFP/Maestro) solution architecture: workspace design, record connections, formula fixes, tier limits, the Planning API, and a Fusion scenario optimizer.
93%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Passed
No findings from the security scan
INDEX.mdskills/workfront-planning-advisory/references/
References index
Map of reference files used by the workfront-planning-advisory skill. Load only what the question requires; do not preload everything.
Top-level synthesis references (read these first)
| File | When to load |
|---|---|
workspace-build-playbook.md | Designing or building a workspace. Categories B and C in SKILL.md routing. |
best-practice-template.md | Referencing the Fréscopa exemplar for patterns OR explicitly checking against the known deviations. |
best-practice-template.json | Trimmed sample export (minified, ~1.6 MB): full structure, ~5 records per record type. Inspect only when a question requires field-level or record-level detail beyond the digest. |
limits-and-tiers.md | Any limits, capacity, sizing, or tier question. Most limits are identical across tiers; check the customer's tier only for records per workspace, total records per instance, or deployment sizing. |
api-contract.md | API v2, versioned filters, PATCH, and bulk operations. |
release-and-access.md | Standalone/bundled licenses, viewing vs editing, field/request permissions, Coworker, dependent connections, business rules, and phased availability. |
public-vs-api-discrepancies.md | Whenever public docs and observed API behavior could disagree (precision, formulas, color palettes, connection naming, identity model, etc.). |
customer-conversation-framings.md | When the user is preparing for or in a customer conversation (limit escalation, P95 ask, RPM comparison, reporting expectations, workspace build engagement, template adoption review). |
Public docs layer (UI/UX surface)
Public Adobe documentation is not bundled. It is fetched live from Experience League, so it never goes stale in this repo.
Use scripts/search.js to find the right pages, then fetch the markdownUrl from the results (any Experience League doc URL with .md appended returns clean markdown):
node scripts/search.js [--all] <keyword1> [keyword2] [...]scripts/docs-index.json backs the search: one entry per Planning documentation page with its title, section, description, headings, and URL. Sections covered: general, architecture, fields, records, views, access, requests, best-practices, api, fusion, ai-assistant, genstudio, canvas-dashboards.
Synthesized references (not available on Experience League)
| File | When to load |
|---|---|
synthesized/automations-deep-dive.md | Canonical decision tree across the 5 automation surfaces (native button-click, native field-change, Fusion, Coworker/retained AI Assistant, request-form approval). Load for Category G. |
synthesized/record-collaboration.md | Comments, history, record layout, and record sharing behavior. |
synthesized/notification-preferences.md | Notification preference behavior. |
API reference layer (programmatic surface)
The API-surface facts most often needed (filter operator sets by field type, formula support gaps, precision limits, connection object types, view behavior) are stated directly in the SKILL.md routing sections, and public-vs-api-discrepancies.md records where public documentation and observed API behavior diverge. api-contract.md adds API v2 guidance and versioned filter examples.
For anything not covered there, confirm against the live Planning API rather than quoting from memory.
Loading strategy
For any given question, the routing in SKILL.md identifies the category (A through N). Load:
- The 1 to 2 top-level synthesis files the category names.
- Any synthesized files explicitly called out.
- If the question needs public UI/UX documentation, run
scripts/search.jsand fetch the top 2 to 3 results. - Stop. Do not preload neighbors.
If a question spans categories, load only the union of files; do not load every reference.
Refresh procedure
Public docs: nothing to refresh. Pages are fetched live from Experience League at answer time. If Adobe publishes new Planning articles, add entries to scripts/docs-index.json (path, url, title, section, description, headings) so the search can surface them.
API-surface facts: the operator sets, precision limits, and function support notes live inline in SKILL.md, with API v2 guidance in api-contract.md. Re-verify them against the live Planning API, or confirm with Adobe, when the product changes.
Best-practice template: re-export when the canonical template changes meaningfully. Update both the .json (raw) and the .md (digest). Re-validate the "Known deviations" section against the current template state.
Discrepancies file: validate the reconciliation table against the live Planning API. Update the date at the bottom of the file when validated.
Date of this index: May 11, 2026.