KubeStellar Console's ACMM dashboard detects AEF governance patterns in any repo

[clubanderson]

Hi — wanted to make you aware that KubeStellar Console now includes an AI-Codebase Maturity Model (ACMM) dashboard that scores any public GitHub repo against a 5-level framework for AI-assisted engineering — and the Agentic Engineering Framework is one of the four source frameworks it detects criteria from.

→ Run your repo through it

What the ACMM dashboard does

The dashboard scans a repo's file structure, workflows, and config for 40+ criteria drawn from four frameworks:

1. ACMM (the 5-level model itself) — our paper at https://arxiv.org/abs/2604.09388
2. Fullsend (fullsend-ai) — readiness + autonomy criteria
3. Agentic Engineering Framework (your project) — governance criteria
4. Claude Reflect (BayramAnnakov) — self-tuning criteria

Each criterion has a detection pattern (file path, workflow name, regex). The dashboard computes a maturity level L1–L5:

Level	Role	Characteristic
L1 Assisted	Executor	AI suggests completions, no persistent instructions
L2 Instructed	Rule-writer	Judgment encoded in CLAUDE.md / AGENTS.md / Copilot instructions
L3 Measured	Analyst	Metrics instrument the AI loop itself
L4 Adaptive	Governor	Metrics feed back into instructions + gating thresholds
L5 Self-Sustaining	Strategist	Codebase proposes/triages/gates its own work

AEF-specific criteria the dashboard detects:

• Task traceability ledger — `.agent/tasks/`, `docs/agent-tasks/`, `agent-tasks.md`
• Structural gates — `CODEOWNERS`, `.agent/boundaries.yml`
• Session continuity doc — CLAUDE.md / AGENTS.md / .cursorrules / copilot-instructions
• Audit trail — `.github/workflows/ai-audit.yml`
• Cross-tool agent config — AGENTS.md / docs/ai-contributors.md
• Change classification policy — `docs/change-classification.md` / `.github/change-tiers.yml`

Why we're reaching out

AEF's governance patterns are the load-bearing signals for the L2–L4 transitions in the ACMM — we treat "code can lie, config cannot" as a core invariant. We wanted you to know your framework is credited as a source and linked from every dashboard scan that detects one of these criteria.

Open source — criteria definitions at:

• AEF criteria set: https://github.com/kubestellar/console/blob/main/web/src/lib/acmm/sources/agentic-engineering-framework.ts?utm_source=github&utm_medium=issue&utm_campaign=acmm_outreach&utm_term=aef
• ACMM paper: https://arxiv.org/abs/2604.09388
• Dashboard source: https://github.com/kubestellar/console/tree/main/web/src/components/acmm?utm_source=github&utm_medium=issue&utm_campaign=acmm_outreach&utm_term=aef

PRs to refine detection patterns are welcome — if AEF defines a signal we're not detecting, we'd like to add it. Feel free to close if not relevant.

[RobertKodes]

Bugs in agent frameworks can be tricky — they often come from async timing, mismatched types, or silent API failures.

Quick debug steps that often work:

1. Check the dev console / logs for the actual error message
2. Try to reproduce with the minimal possible input
3. If it's an API issue, check if the provider changed their response format
4. Add console.log / print statements at each step to isolate where it breaks

If you can share the error trace, I might be able to help. Good luck tracking it down!