feat: polish UI demo workflow

README.md

@@ -337,11 +337,14 @@ Refs:
 
 `cfg ui` starts a localhost-only web UI over the same action layer as the CLI and
 MCP server. It reads like a git client: a collection-and-record tree on the left,
-a commit-graph history rail, and a line-aligned side-by-side diff that collapses
-unchanged context (expandable in place) and keeps the field name pinned while you
-scroll. It can run status, diff, impact, commit, branch draft commits, PR open
-and merge, log, show, adopt, restore, tag, init, import, and fsck, and ships
-dark and light themes.
+a recent-activity rail before you select anything, per-record commit graphs, and
+a line-aligned side-by-side diff that collapses unchanged context (expandable in
+place) and keeps the field name pinned while you scroll. The recent activity view
+surfaces live drift and the latest cfgit commits across all configured records,
+so you can see what changed recently without opening records one by one. It can
+run status, diff, impact, commit, branch draft commits, PR open and merge, log,
+show, adopt, restore, tag, init, import, and fsck, and ships dark and light
+themes.
 
 By default it binds to `127.0.0.1:8765` and tries the next free ports if needed:
 
@@ -354,6 +357,35 @@ If you omit `--port`, cfgit will try the next free local ports. If you pass
 `--port` explicitly, cfgit treats that port as intentional and fails if it is
 already in use.
 
+## Synthetic UI demo
+
+The repo includes a fake support-control-plane fixture for screenshots, demos,
+and UI testing. It creates only synthetic records in the database you pass; use a
+throwaway database name when you want to keep the run contained.
+
+```bash
+python examples/seed_support_demo.py \
+  --uri 'mongodb://localhost:27017/?replicaSet=rs0' \
+  --db cfgit_ui_demo \
+  --reset
+
+cfg --config-file examples/cfgit-support-demo.toml init
+cfg --config-file examples/cfgit-support-demo.toml import --all -m "initial synthetic demo import"
+
+python examples/seed_support_demo.py \
+  --uri 'mongodb://localhost:27017/?replicaSet=rs0' \
+  --db cfgit_ui_demo \
+  --drift
+
+cfg --config-file examples/cfgit-support-demo.toml ui
+```
+
+The first seed creates clean planner, critic, router, model, and policy records.
+The `--drift` pass then simulates an admin-console edit: planner routing changes,
+the refund policy changes, and a new entitlements resolver appears. That gives
+the UI useful drift, impact, adopt, branch, PR, merge, and restore-history paths
+without using proprietary data.
+
 ## MCP and agent usage
 
 The MCP server exposes the same operations with a uniform envelope:
@@ -385,6 +417,7 @@ Tools include:
 - `cfg_pr_show`
 - `cfg_pr_close`
 - `cfg_pr_merge`
+- `cfg_recent_history`
 - `cfg_log`
 - `cfg_show`
 - `cfg_adopt`

docs/USAGE.md

@@ -269,7 +269,10 @@ cfg ui
 ```
 
 The UI binds to `127.0.0.1`. It is a local operator surface over the same action
-layer as the CLI and MCP server.
+layer as the CLI and MCP server. Before a record is selected, the middle rail
+shows recent activity across all configured records: current live drift plus the
+latest cfgit history entries. Selecting one of those entries opens that record's
+normal history and diff view.
 
 ## JSON mode
 

examples/cfgit-support-demo.toml

@@ -0,0 +1,67 @@
+[project]
+name = "cfgit-support-demo"
+
+[history]
+history_collection = "cfgit_demo_history"
+heads_collection = "cfgit_demo_heads"
+
+[branches]
+enabled = true
+refs_collection = "cfgit_demo_refs"
+default_branch = "main"
+
+[[collection]]
+name = "agent_configs"
+id_field = "config_id"
+live_when = { is_active = true }
+ignore_fields = ["_id", "is_active", "updated_at", "updated_by"]
+secret_fields = []
+
+[[collection]]
+name = "modelgarden_models"
+id_field = "model_path"
+live_when = {}
+ignore_fields = ["_id", "updated_at", "updated_by"]
+secret_fields = ["provider_config.api_key"]
+
+[[collection]]
+name = "policy_rules"
+id_field = "rule_id"
+live_when = { active = true }
+ignore_fields = ["_id", "active", "updated_at", "updated_by"]
+secret_fields = []
+
+[secrets]
+block_fields = ["*_key", "*_secret", "*_token", "*api_key*", "*password*"]
+block_values = ["sk-[A-Za-z0-9]{20,}", "AKIA[0-9A-Z]{16}"]
+on_match = "refuse"
+
+[author]
+from = "git"
+
+[connections]
+enabled = true
+share_with_ai = []
+ai_provider = "openai"
+warn_level = "none"
+links = [
+  { field = "phase_contract", means = "contract other records may rely on" },
+  { field = "tools", means = "shared tool list" },
+  { field = "fallback_models", means = "fallback model routing" },
+  { field = "applies_to", means = "policy-to-agent linkage" },
+]
+
+[env.dev]
+database = "mongo"
+uri = "mongodb://localhost:27017/?replicaSet=rs0"
+db = "cfgit_ui_demo"
+needs_approval = false
+
+[env.dev.identity]
+mode = "open"
+
+[env.dev.permissions]
+mode = "open"
+admins = []
+writers = []
+admin_actions = ["restore_system"]

examples/seed_support_demo.py

@@ -0,0 +1,228 @@
+"""Seed a synthetic cfgit demo database.
+
+This fixture is intentionally fake: support agents, model routes, and policy
+rules for demo screenshots. It never reads production data.
+
+Typical demo flow:
+  python examples/seed_support_demo.py --reset
+  cfg --config-file examples/cfgit-support-demo.toml init
+  cfg --config-file examples/cfgit-support-demo.toml import --all -m "initial import"
+  python examples/seed_support_demo.py --drift
+"""
+from __future__ import annotations
+
+import argparse
+from datetime import datetime, timezone
+
+from pymongo import MongoClient
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--uri", default="mongodb://localhost:27017/?replicaSet=rs0")
+    parser.add_argument("--db", default="cfgit_ui_demo")
+    parser.add_argument("--reset", action="store_true")
+    parser.add_argument(
+        "--drift",
+        action="store_true",
+        help="apply synthetic out-of-band edits after the base records have been imported",
+    )
+    args = parser.parse_args()
+
+    client = MongoClient(args.uri)
+    db = client[args.db]
+    if args.db in {"admin", "config", "local"}:
+        raise SystemExit(f"refusing to seed Mongo system database {args.db!r}")
+    if args.reset:
+        for name in [
+            "agent_configs",
+            "modelgarden_models",
+            "policy_rules",
+            "cfgit_demo_history",
+            "cfgit_demo_heads",
+            "cfgit_demo_refs",
+        ]:
+            db.drop_collection(name)
+
+    now = datetime.now(timezone.utc)
+    if args.drift:
+        apply_drift(db, now)
+        print(f"applied synthetic cfgit demo drift in Mongo database {args.db!r}")
+        return
+
+    seed_base(db, now)
+    print(f"seeded synthetic cfgit demo data in Mongo database {args.db!r}")
+
+
+def seed_base(db, now: datetime) -> None:
+    db.agent_configs.insert_many(
+        [
+            {
+                "config_id": "planner",
+                "is_active": True,
+                "role": "Support Triage Planner",
+                "model": "openai/gpt-4o-mini",
+                "tools": ["search", "calendar", "handoff"],
+                "fallback_models": ["anthropic/claude-haiku"],
+                "phase_contract": (
+                    "Produces a support-ticket plan with risk, refund-policy, "
+                    "and owner handoff notes."
+                ),
+                "instructions": (
+                    "Classify incoming support tickets, cite the current refund "
+                    "policy, and hand off risky cases."
+                ),
+                "updated_at": now,
+                "updated_by": "seed",
+            },
+            {
+                "config_id": "critic",
+                "is_active": True,
+                "role": "Policy Reviewer",
+                "model": "anthropic/claude-haiku",
+                "tools": ["diff", "policy_check"],
+                "phase_contract": (
+                    "Reviews planner decisions for refund-policy compliance and "
+                    "escalation risk."
+                ),
+                "instructions": (
+                    "Review planner output from planner, verify refund_window_v1, "
+                    "and flag missing restore notes."
+                ),
+                "updated_at": now,
+                "updated_by": "seed",
+            },
+            {
+                "config_id": "router",
+                "is_active": True,
+                "role": "Support Router",
+                "model": "openai/gpt-4o-mini",
+                "tools": ["modelgarden"],
+                "fallback_models": ["openai/gpt-4o-mini"],
+                "phase_contract": (
+                    "Routes customer tickets to planner or critic according to "
+                    "policy risk."
+                ),
+                "instructions": (
+                    "Use planner for standard tickets and critic when "
+                    "refund_window_v1 or escalation rules apply."
+                ),
+                "updated_at": now,
+                "updated_by": "seed",
+            },
+        ]
+    )
+    db.modelgarden_models.insert_many(
+        [
+            {
+                "model_path": "openai/gpt-4o-mini",
+                "provider": "openai",
+                "enabled": True,
+                "retry_policy": "standard",
+                "price_per_million": 0.15,
+                "provider_config": {"api_key": "demo-secret-stays-live"},
+            },
+            {
+                "model_path": "anthropic/claude-haiku",
+                "provider": "anthropic",
+                "enabled": True,
+                "retry_policy": "conservative",
+                "price_per_million": 0.8,
+                "provider_config": {"api_key": "demo-secret-stays-live"},
+            },
+        ]
+    )
+    db.policy_rules.insert_many(
+        [
+            {
+                "rule_id": "refund_window_v1",
+                "active": True,
+                "title": "Refund window",
+                "applies_to": ["planner", "critic", "router"],
+                "severity": "medium",
+                "rule_text": (
+                    "Refunds are allowed within 30 days unless the ticket is "
+                    "marked abuse_risk."
+                ),
+                "updated_at": now,
+                "updated_by": "seed",
+            },
+            {
+                "rule_id": "abuse_risk_v1",
+                "active": True,
+                "title": "Abuse-risk escalation",
+                "applies_to": ["critic", "router"],
+                "severity": "high",
+                "rule_text": (
+                    "Cases with repeated refund attempts require human review "
+                    "before resolution."
+                ),
+                "updated_at": now,
+                "updated_by": "seed",
+            },
+        ]
+    )
+
+
+def apply_drift(db, now: datetime) -> None:
+    db.agent_configs.update_one(
+        {"config_id": "planner"},
+        {
+            "$set": {
+                "model": "openai/gpt-4o-mini-2026-demo",
+                "fallback_models": ["anthropic/claude-haiku", "openai/gpt-4o-mini"],
+                "phase_contract": (
+                    "Produces a support-ticket plan with refund policy, entitlement checks, "
+                    "and escalation owner notes."
+                ),
+                "instructions": (
+                    "Classify incoming support tickets, cite refund_window_v1, check "
+                    "entitlement tier, and hand off risky cases."
+                ),
+                "updated_at": now,
+                "updated_by": "demo-drift",
+            }
+        },
+    )
+    db.policy_rules.update_one(
+        {"rule_id": "refund_window_v1"},
+        {
+            "$set": {
+                "rule_text": (
+                    "Refunds are allowed within 45 days unless the ticket is marked "
+                    "abuse_risk or enterprise_contract_override."
+                ),
+                "severity": "high",
+                "updated_at": now,
+                "updated_by": "demo-drift",
+            }
+        },
+    )
+    db.agent_configs.update_one(
+        {"config_id": "entitlements"},
+        {
+            "$set": {
+                "config_id": "entitlements",
+                "is_active": True,
+                "role": "Entitlements Resolver",
+                "model": "openai/gpt-4o-mini",
+                "tools": ["search", "billing", "handoff"],
+                "fallback_models": ["anthropic/claude-haiku"],
+                "phase_contract": (
+                    "Checks plan, renewal date, and regional refund obligations before "
+                    "planner response."
+                ),
+                "instructions": (
+                    "Resolve customer entitlement tier and return constraints before "
+                    "refunds are discussed."
+                ),
+                "updated_at": now,
+                "updated_by": "demo-drift",
+            }
+        },
+        upsert=True,
+    )
+
+
+if __name__ == "__main__":
+    main()

skills/cfgit/SKILL.md

@@ -33,6 +33,7 @@ Use cfgit as the safety layer around a live datastore. The app still reads and w
 2. Inspect before mutation.
    - `cfg diff <record> =HEAD =live --json`
    - `cfg log <record> --json`
+   - Use MCP `cfg_recent_history` when you need the latest cfgit commits across all configured records before choosing a record.
    - `cfg impact <record> =HEAD =live --json`
    - Scoped LLM review, when explicitly needed: `cfg impact <record> =HEAD =live --against <related-record> --llm --json`
    - If history lookup reports an env mismatch, re-run against the stamped env before making changes.
@@ -83,6 +84,7 @@ If the cfgit MCP server is available, prefer its tools over shelling out:
 - `cfg_pr_show`
 - `cfg_pr_close`
 - `cfg_pr_merge`
+- `cfg_recent_history`
 - `cfg_log`
 - `cfg_show`
 - `cfg_adopt`