continuous-improvement-loop — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited continuous-improvement-loop (Agent Skill) and scored it 96/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 1 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Part 12 is the continuous improvement loop that runs alongside live operations from go-live onwards. It aggregates market signals and operating signals into recommendations that feed back into the brand's product, offering, and service decisions.
Heavy skill. Grep before Read any referenced file, then Read only matched ranges with offset + limit. List ${CLAUDE_PLUGIN_DATA}/<brand>/ before opening files. On re-invocation mid-session, skip files already in context.
This is not a one-time activity. It runs perpetually once Part 11 is complete, with formal output at each Quarterly Business Review (QBR) and ad-hoc output when significant signals warrant.
Without an explicit feedback loop, marketing operates on assumptions made months ago. Markets shift, customers evolve, competitors move, products are refined — but if these shifts do not flow back into the strategy, the engagement silently grows stale.
Part 12 closes the loop:
Every quarterly review (per reporting-cadence.md) generates structured signals:
Feedback from across customer touchpoints:
From the ongoing competitor monitoring (existing /digital-marketing-pro:competitor-monitor skill):
Insights from execution that the team surfaces:
Part 12 is active continuously, with structured outputs:
| Cadence | Trigger | Output |
|---|---|---|
| Daily / weekly | Automated signal capture as part of normal operations | Signals logged to part-12-continuous-improvement/signals.jsonl |
| Monthly | Monthly performance report | "Signals This Month" section in the report; logged to signals.jsonl |
| Quarterly | QBR | Structured Part 12 deliverable — see below |
| Ad-hoc | Significant signal (e.g., competitor product shift, sales team flagging recurring objection, KPI suddenly cratering) | Ad-hoc Part 12 brief produced within 1 week |
Each quarter, the continuous loop produces a structured deliverable for the brand business owners — not just marketing leadership.
---
document: part-12-quarterly-improvement-brief
engagement: {engagement-id}
quarter: {YYYY-Qn}
produced: {iso-timestamp}
audience: brand business leadership
---
# Quarterly Product & Offering Improvement Brief — {Quarter}
## Executive Summary
(3-5 sentences. The signals that matter most. The recommendations that follow.)
## Signal Aggregation
### Market signals
{Macro market shifts observed in the quarter}
### Customer signals
{Aggregated themes from customer feedback, ORM, sales conversations}
### Competitive signals
{Competitor moves that warrant response or reflection}
### Operating signals
{Patterns from execution — campaigns that under/outperformed; audience surprises; channel shifts}
## Implications
### For the brand strategy
{What in the v2 strategy looks confirmed by the quarter? What looks weakened? Anything that warrants v2.x update-back?}
### For the channel mix
{Any channel reweighting recommended?}
### For the product / offering
{This is the unique Part 12 contribution. What signals suggest the product or offering itself should change?}
## Recommendations
### To the marketing team
{Tactical adjustments — typically already in flight from monthly optimisation, but formalised here}
### To the product / business team
{The substantive Part 12 output — recommendations about product, offering, pricing, distribution that flow from marketing's vantage point}
### To leadership
{Strategic considerations that span functions}
## Triggers for v2.x Update-Back
(If any of the signals warrant a source-document version bump per the [update-back-rule.md](../context-engine/update-back-rule.md), list them here. The actual update-back happens via /digital-marketing-pro:engagement update-back.)
## Open Questions Raised This Quarter
(Things the data raises but cannot answer without further investigation.)engagements/{id}/part-12-continuous-improvement/quarterly-briefs/{YYYY-Qn}-quarterly-improvement-brief.mdPlus PDF export for distribution to leadership.
When a significant signal lands between QBRs, the loop produces an ad-hoc brief:
Ad-hoc briefs are short (1–3 pages), fast (within a week of the signal), and action-oriented (recommend a specific response).
Output location:
engagements/{id}/part-12-continuous-improvement/ad-hoc-briefs/{YYYY-MM-DD}-{slug}.mdsignals.jsonl for the quarterquarterly-briefs/ad-hoc-briefs/The plugin captures signals continuously via:
/digital-marketing-pro:engagement signal --source <source> --signal "<description>" (custom command — uses the engagement-state.py lif-log-change pattern)All signals append to signals.jsonl:
{"timestamp":"...","source":"customer_feedback","signal":"3 sales reps reported customers asking for X integration","severity":"medium"}
{"timestamp":"...","source":"competitor_monitor","signal":"Competitor Y launched freemium tier","severity":"high"}
{"timestamp":"...","source":"performance_anomaly","signal":"LinkedIn CPL dropped 35% week over week","severity":"high","investigate":true}/digital-marketing-pro:engagement update-back command is invoked separately after explicit approval.User: Q2 ended last week. Produce the quarterly improvement brief.
Skill response:
1. Grep first across the Q2 monthly reports (April, May, June) for the metrics you need; targeted Read with offset+limit on the matching lines — do not bulk-Read all three reports
2. Stream signals.jsonl line-by-line for Q2 (87 signals) — use head/tail/grep, not full Read
3. Grep the competitor monitor outputs (3 significant changes); targeted Read on the matching findings
4. Grep customer feedback aggregation (47 reviews + 12 sales-rep notes) for the themes you need; targeted Read on the matches
5. Aggregate into 4 signal categories
6. Identify implications:
- Strategy: v2 positioning held up — confirmed by 12 customer testimonials echoing the positioning language
- Channels: LinkedIn over-performed (CPL 30% below moderate scenario); Meta under-performed (CPL 40% above)
- Product: 8 customer signals + 4 sales-rep notes suggest Feature X is consistently asked for
7. Recommendations:
- Marketing: shift 25% of Meta budget to LinkedIn next quarter
- Product: prioritise Feature X for Q3 roadmap (8 named customer asks)
- Leadership: consider a freemium tier (competitor moved here; 5 sales-rep notes about price-sensitive prospects)
8. Identify update-back triggers: Core Doc 3.4 (channel mix) needs v2.1 to reflect LinkedIn's higher allocation
9. Save quarterly-briefs/2026-Q2-quarterly-improvement-brief.md (~8 pages)
10. Brief: "Q2 Improvement Brief produced. 1 update-back trigger flagged. Run /digital-marketing-pro:engagement update-back --doc 3.4 --reason 'Q2 evidence: LinkedIn outperformed; Meta underperformed; recommend channel reweight' after leadership approval."User: Three customers in the past two weeks have switched to a competitor that just launched a freemium tier. We need a fast read on this.
Skill response:
1. Confirm significance with engagement owner ✓
2. Read inputs: the 3 churn cases, competitor monitor on the competitor's freemium launch, last 90 days of churn data for pattern check
3. Draft brief:
- Signal: 3 churns to Competitor Y in 14 days; pattern check shows churn rate to Y up 4x vs prior 90 days
- Evidence: churn interview notes (2 of 3 cited price); Competitor Y launched freemium 2026-04-15
- Implications: short-term — defensive offer for at-risk segment; long-term — pricing strategy review warranted
- Recommended response: (1) marketing — defensive offer to current at-risk customers within 7 days; (2) product/leadership — assess freemium response within 30 days
- Decision deadline: response plan by 2026-05-12
4. Save ad-hoc-briefs/2026-05-05-competitor-y-freemium-response.md (2 pages)
5. Distribute per approval chainengagement-workflow — engagement orchestrationcompetitor-monitor, performance-monitor-agent, intelligence-curator, quality-assurance~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.