You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
docs(v1.6.1): playbook freshness pass — lead with af workforce init
The v1.6.0 release shipped `af workforce init` as the native deploy default
but three onboarding playbooks still pointed users at the deprecated
`af paperclip init` — a real UX regression caught by the blueprint +
playbook verification pass.
Updated:
- `first-touch`: "Deploy a team to AgenticFlow (native, v1.6+)" section
now leads; paperclip section demoted to "DEPRECATED — sunset 2026-10-14"
- `amazon-seller`: full rewrite with OPTION A (native) → OPTION B
(paperclip deprecated). Shows `--include-optional-slots` for the full
5-agent team, MCP-attach-per-agent recipe pointing at the
mcp-client-quirks playbook
- `company-from-scratch`: flow restructured so native workforce comes
first; paperclip path retained as legacy with clear sunset
- `deploy-to-paperclip`: prominent ⚠️ DEPRECATED header + pointers to
migrate-from-paperclip and amazon-seller/company-from-scratch playbooks
Verified live:
- All 6 blueprints (dev-shop, marketing-agency, sales-team, content-studio,
support-center, amazon-seller) translate cleanly through --dry-run
- Full e2e `content-studio` deploy: creates workforce + 3 agents (ceo +
cmo + engineer), wired graph validates with 0 errors, graph shape
confirmed trigger → ceo → {cmo, engineer, output}
- Clean delete of all 4 resources (workforce + 3 agents) via the
v1.5.2 envelope
Tests: 392 pass, 14 skipped, 10 todo — unchanged (content-only patch).
"Playbook freshness pass — three onboarding playbooks (`first-touch`, `amazon-seller`, `company-from-scratch`) now lead with `af workforce init` (v1.6 native deploy) and show the deprecated `af paperclip init` path as a secondary option with a sunset date",
21
+
"`deploy-to-paperclip` playbook got a prominent `⚠️ DEPRECATED` header pointing at `af playbook migrate-from-paperclip` and the sunset 2026-10-14",
22
+
"`amazon-seller` playbook expanded: shows `--include-optional-slots` to create all 5 agents, attach-MCP-per-agent recipe pointing at `mcp-client-quirks`, and smoke-run/publish commands",
23
+
"Verified: all 6 blueprints translate cleanly through `--dry-run` (node/edge math correct). Full e2e `content-studio` deploy creates + wires + validates + tears down cleanly",
24
+
],
25
+
for_ai: [
26
+
"When `af bootstrap --json` returns workforces=0 and the user wants a team, the FIRST thing to show them is `af workforce init --blueprint <id> --dry-run --json` — the playbooks now lead with this, no longer with paperclip",
27
+
"For Amazon Singapore sellers, run `af playbook amazon-seller` — it shows the one-command native deploy path first, with `--include-optional-slots` for the full 5-agent team",
28
+
"`af playbook deploy-to-paperclip` now explicitly flags itself deprecated and points at `migrate-from-paperclip` — follow that path unless the user has an existing Paperclip instance",
summary: "Publish AgenticFlow agents to a Paperclip company where they receive tasks, execute work, and report results — all fromthe CLI.",
339
+
title: "Deploy Agents to Paperclip (DEPRECATED)",
340
+
summary: "[DEPRECATED — sunset 2026-10-14] Publish AgenticFlow agents to a Paperclip company. Use `af playbook migrate-from-paperclip` for the AgenticFlow-native alternative.",
330
341
content: [
342
+
"# ⚠️ DEPRECATED — Migrate to `af workforce` by 2026-10-14",
343
+
"",
344
+
"This playbook describes the legacy Paperclip-based deploy. As of CLI v1.5.0,",
"VERIFY: `status: \"ok\"`. Replace PAPERCLIP_URL with the user's Paperclip URL (default: http://localhost:3100).",
519
-
"IF FAIL: Paperclip is not running. Ask the user to start it.",
520
-
"",
521
-
"```bash",
522
544
"af agent list --fields id,name,model --json",
523
545
"```",
524
-
"VERIFY: at least 1 agent exists. These are the agents that will work in the company.",
525
-
"IF EMPTY: the user needs to create agents first — run `af playbook quickstart`.",
546
+
"VERIFY: at least 1 agent exists in the workspace (not strictly required — `af workforce init` creates fresh ones). If you want to re-use existing agents, note their IDs.",
summary: "Deploy a 5-agent AI team for Amazon Singapore sellers — listing optimization, PPC, competitor analysis, pricing, and customer support. One command setup.",
690
+
summary: "Deploy a 5-agent AI team for Amazon Singapore sellers — listing optimization, PPC, competitor analysis, pricing, and customer support. One command.",
636
691
content: [
637
692
"Deploy an AI team that runs your Amazon Singapore business 24/7.",
"3. Trigger agents via `/webhook/paperclip` with issueId from task list.",
716
-
"4. Each agent responds with structured analysis posted as issue comments.",
717
-
"5. Use `af agent run --agent-id <AF_ID> --message <task> --json` for direct queries.",
768
+
"1. Default: `af workforce init --blueprint amazon-seller --name \"<name>\" --include-optional-slots --json` creates a runnable 5-agent team in one call.",
769
+
"2. Attach MCPs/tools to each agent via `af agent update --patch` (tools start empty on init).",
770
+
"3. Smoke test with `af workforce run --workforce-id <id> --trigger-data '{\"message\":\"...\"}'`.",
771
+
"4. Publish a public URL via `af workforce publish --workforce-id <id>` when ready to hand off.",
772
+
"5. Clean up test deploys with `af workforce delete --workforce-id <id>` + `af agent delete --agent-id <id>` for each created agent.",
0 commit comments