Agentic Workflow Portfolio Yield Report — 2026-05-26

[github-actions]

Agentic Workflow Portfolio Yield Report

Date: 2026-05-26 | Workflows analyzed: 236 | Evidence quality: Low (telemetry backend unavailable at precompute time; Sentry live data used for validation)

---

Executive Summary

The gh-aw workflow portfolio of 236 workflows has a computed yield of −89.12, driven by extreme overlap drag (356.57). The portfolio operates at 97.2% agentic fraction with only 2.8% deterministic logic — near-zero guardrails across most workflows. Fragmentation is maximal (1.0) and governance drag is critically high (0.91).

The precompute reported 0% telemetry coverage, but live Sentry validation contradicts this: ~24 workflows have active spans in the past 7 days, confirming the Grafana telemetry backend was unavailable during precompute rather than workflows lacking instrumentation. Sentry identifies Smoke Copilot as the highest-failure workflow (64 failures, 16.3% failure rate) and Deployment Incident Monitor as a degraded operational workflow (6 failures, strict mode disabled).

The portfolio's central structural problem is not individual workflow quality but fragmentation: 189 of 236 workflows are merge candidates due to overlapping semantic scope, and only 3 meet the bar for unconditional keep.

---

Portfolio Health

Metric	Value	Signal
Workflow count	236	🔴 Oversized
Portfolio yield	−89.12	🔴 Critical
Overlap drag	356.57	🔴 Critical
Average agentic fraction	97.2%	🔴 Near-zero determinism
Maintenance drag	0.89	🔴 High
Portfolio cost	0.60	🟡 Elevated
Portfolio risk	0.67	🟡 Elevated
Fragmentation	1.0	🔴 Maximum
Governance drag	0.91	🔴 Critical
Reuse signal	0.89	🟡 Mixed (high but untested)
Trust concentration	0.24	🔴 Low
Telemetry observed (Sentry, 7d)	~24 workflows	🟡 Partial

---

Workflow Portfolio

Summary by recommendation category:

Category	Count	Rationale
✅ Keep	3	High yield, low overlap, acceptable risk
✏️ Revise	9	Recoverable with targeted hardening
🔀 Merge	189	Overlap-driven consolidation candidates
📡 Instrument	29	Missing telemetry or safe-outputs
🗑️ Retire	6	Risk ≥ 0.80, all compliance checks failing

Keep

Workflow	Notes
aw-portfolio-yield.md	Core governance infrastructure; lowest agentic fraction (0.374)
example-permissions-warning.md	Canonical permissions reference; fix: enable strict mode
terminal-stylist.md	Unique utility with no overlap peers; fix: add safe-outputs

Retire

Workflow	Risk	Critical Issues
ace-editor.md	1.0	No strict mode, no timeout, no safe-outputs
copilot-pr-nlp-analysis.md	1.0	Stale lockfile, no strict mode, no safe-outputs
daily-semgrep-scan.md	0.80	No strict mode, no timeout (security workflow)
daily-sentrux-report.md	0.94	No strict mode, no timeout, no safe-outputs
firewall.md	0.98	Stale lockfile, no strict mode, no safe-outputs
test-workflow.md	1.0	No strict mode, no timeout, no safe-outputs, no value

Top Revise Candidates

Workflow	Risk	Issues	Sentry
deployment-incident-monitor.md	0.73	No strict mode	6 failures/7d 🔴
constraint-solving-potd.md	0.84	No strict mode, no timeout	2 failures/7d
daily-model-inventory.md	0.68	Stale lockfile, highest cost (0.675)	—
weekly-editors-health-check.md	0.41	High cost (0.672), no telemetry	—
dependabot-repair.md	0.81	No strict mode, no timeout	—

---

Overlap Clusters

Three high-priority consolidation clusters identified (overlap > 0.75):

Cluster 1 — OTel Instrumentation Advisors (overlap: 0.978) 🔴 Near-duplicate

daily-grafana-otel-instrumentation-advisor.md  <─┐
daily-otel-instrumentation-advisor.md          ───┘  Merge: Grafana as optional backend flag

Cluster 2 — Smoke Agent Scope Tests (overlap: 0.787)

smoke-agent-all-merged.md    <─┐
smoke-agent-all-none.md      ──┤  Merge: parameterize scope (all/public, merged/none)
smoke-agent-public-none.md   ──┘

Cluster 3 — Engine Smoke Tests (overlap: 0.745)

smoke-antigravity.md  <─┐
smoke-gemini.md       ──┤  Merge: engine matrix workflow
smoke-pi.md           ──┘
  + smoke-opencode.md, smoke-crush.md, smoke-claude.md, smoke-copilot-arm.md, smoke-codex.md

Additional high-overlap pairs (0.60–0.75):

Pair	Overlap	Action
smoke-create-cross-repo-pr + smoke-update-cross-repo-pr	0.682	Merge CRUD pair
research + scout	0.655	Differentiate or merge
daily-safe-output-optimizer + safe-output-health	0.635	Merge with mode flag
copilot-pr-prompt-analysis + prompt-clustering-analysis	0.632	Merge
grumpy-reviewer + pr-code-quality-reviewer	0.622	Retire grumpy into active PR reviewer
dependabot-campaign + dependabot-worker	0.639	Evaluate orchestrator/worker split

---

Episode-Level Observations

• No episode_metrics were computed in precompute — cross-workflow coordination patterns are not yet tracked.
• The smoke-engine cluster (antigravity, gemini, pi, opencode, crush, claude, copilot-arm, codex) spans 8 workflows testing different AI engines with near-identical structure — a natural episode candidate for a single parameterized smoke suite.
• The daily-* prefix group (30+ workflows) shows no episode-level coordination despite sharing daily cadence — potential for a unified daily maintenance episode with shared health reporting.
• Sentry data shows Smoke CI (1970 spans) and Test Quality Sentinel (991 spans) running at similar cadences — these may form a natural CI quality episode.

---

Organizational Health Signals

Signal	Score	Interpretation
Fragmentation	1.00	Maximum — every workflow is a silo
Governance drag	0.91	Critically elevated by broad scope + missing telemetry
Reuse signal	0.89	High declared reuse, unconfirmed (telemetry backend gap)
Trust concentration	0.24	Low — execution trust spread thinly across many low-confidence workflows

The portfolio exhibits a classic sprawl pattern: rapid workflow creation without consolidation, no episode-level coordination, and telemetry gaps masking actual value delivery.

---

Deterministic vs Agentic Findings

• 97.2% portfolio agentic fraction — near-zero deterministic validation across 236 workflows
• Only aw-portfolio-yield.md achieves a meaningfully low agentic fraction (0.374) with real pre/post-agent guardrails
• All 6 retirement candidates have agentic_fraction = 1.0 and zero deterministic steps
• Revise candidates average risk 0.628 vs retire candidates 0.954 — revise group is recoverable
• The precompute telemetry_coverage = 0.0 is a Grafana backend availability gap — Sentry live data confirms real telemetry for ~24 workflows

---

Highest-Value Actions

1. 🔀 Merge daily-grafana-otel-instrumentation-advisor + daily-otel-instrumentation-advisor — 0.978 overlap, near-duplicate. Make Grafana an optional backend parameter.

2. 🔀 Parameterize the smoke-agent-all-* cluster into one matrix workflow with scope flags — all 3 are confirmed active, 0.787 overlap.

3. 🗑️ Retire firewall.md, ace-editor.md, test-workflow.md — risk ≥ 0.981, all compliance checks failing, zero yield.

4. 📡 Instrument daily-compiler-quality.md and deployment-incident-monitor.md — both have confirmed Sentry failures but insufficient telemetry to diagnose.

5. 🔧 Fix deployment-incident-monitor.md strict mode — 6 Sentry failures in 7 days with strict mode disabled. Highest operational risk in production workflows.

---

Retirement Candidates

Workflow	Risk	Blockers to Retirement
ace-editor.md	1.0	None — no consumers observed
copilot-pr-nlp-analysis.md	1.0	Verify not referenced by other workflows
daily-semgrep-scan.md	0.80	Confirm security scanning covered elsewhere
daily-sentrux-report.md	0.94	Archive last report output before retiring
firewall.md	0.98	Stale lockfile suggests already inactive
test-workflow.md	1.0	None

---

Consolidation Opportunities

Priority	Workflows	Overlap	Estimated Reduction
🔴 Critical	grafana-otel + otel-advisor	0.978	2 → 1
🔴 Critical	smoke-agent-all-* (3 workflows)	0.787	3 → 1
🟡 High	smoke engine cluster (8 workflows)	0.65–0.745	8 → 1 matrix
🟡 High	smoke CRUD pair	0.682	2 → 1
🟡 High	research + scout	0.655	2 → 1
🟢 Medium	safe-output-optimizer + safe-output-health	0.635	2 → 1
🟢 Medium	copilot-pr-prompt + prompt-clustering	0.632	2 → 1
🟢 Medium	grumpy-reviewer → pr-code-quality-reviewer	0.622	2 → 1

Potential portfolio reduction: 236 → ~150 workflows if all Tier 1–2 consolidations are executed.

---

Instrumentation Gaps

1. Precompute telemetry_coverage = 0.0 is a backend gap, not a true gap — Sentry confirms ~24 active workflows. Future precomputes should fall back to Sentry when Grafana is unavailable.

2. gh-aw.run.status null for ~50% of Sentry spans — status field population in send_otlp_span.cjs is inconsistently invoked, masking half of all failure rates.

3. Zero Sentry error events — gh-aw workflows only emit span-level gh-aw.run.status:failure, never Sentry error events. This prevents alerting and issue auto-creation.

4. 29 workflows have no OTEL spans. Priority targets:

   • agentic-token-audit.md — has safe-outputs, missing OTEL
   • daily-agentrx-trace-optimizer.md — ironic gap for a trace optimizer
   • daily-compiler-quality.md — 4 confirmed Sentry failures need context
   • copilot-cli-deep-research.md — research outcomes not tracked

---

Deterministic Portfolio JSON

Click to expand precompute summary

{
  "workflow_count": 236,
  "portfolio_yield": -89.1218,
  "portfolio_cost": 0.6005,
  "portfolio_risk": 0.6653,
  "portfolio_maintenance_drag": 0.8877,
  "portfolio_overlap_drag": 356.5721,
  "average_agentic_fraction": 0.9716,
  "average_deterministic_fraction": 0.0284,
  "observability_declared_coverage": 0.8771,
  "telemetry_coverage": 0.0,
  "evidence_quality": "low",
  "organizational_health": {
    "fragmentation": 1.0,
    "governance_drag": 0.9081,
    "reuse": 0.8941,
    "trust_concentration": 0.2425
  },
  "recommendations_seed": {
    "keep": 3,
    "revise": 9,
    "merge": 189,
    "instrument": 29,
    "retire": 6
  },
  "top_overlap_cluster": {
    "workflows": [
      "daily-grafana-otel-instrumentation-advisor.md",
      "daily-otel-instrumentation-advisor.md"
    ],
    "max_overlap": 0.9777
  },
  "sentry_validation": {
    "active_workflows_7d": "~24",
    "highest_failure_rate": "Smoke Copilot: 64 failures, 16.3%",
    "highest_operational_failures": "Deployment Incident Monitor: 6 failures/7d",
    "portfolio_yield_run_failures": 4,
    "run.status_null_rate": "~50% of spans"
  }
}

Generated by 📊 Agentic Workflow Portfolio Yield · sonnet46 1.9M · ◷

• expires on Jun 25, 2026, 5:02 PM UTC