Status: scaffolded Publication state: public scaffold Release state: not released
agentic-engineering-workforce is a Phase 7 public scaffold for public-safe agentic engineering lifecycle discipline under human authority.
This repo is not an active agent workforce, autonomous production agent system, internal prompt library, private agent memory store, live tool token record, production execution log, private runtime map, or released software package.
| Field | Value |
|---|---|
| Status | scaffolded |
| Publication state | public scaffold |
| Release state | not released |
| Public GitHub repo | published scaffold |
| First artifact | execution-loops/codex-style-execution-loop.md completed as scaffolded synthetic artifact |
| Published artifact | scaffolded |
| Released artifact | private/not-public or not released |
| Profile routing | planned |
| Proof-stack routing | planned |
Agents assist with bounded tasks under human review. Humans remain architectural authority for creation, publication, metadata, routing, release, refusal overrides, and any boundary-bearing decision.
Agent examples are synthetic role and lifecycle patterns only. They do not imply autonomous production agents, active agent workforce deployment, or production execution authority.
Execution loops cover intake, planning, scoped patching, validation, handoff, refusal, and human approval.
Specialist roles are public-safe role cards. They do not include internal specialist prompts, private memory, private runtime details, or internal coordination internals.
Tool permissions are synthetic policy examples. No live tool tokens, credentials, private repo paths, or production endpoints may appear.
Review gates define human approval before creation, publication, metadata changes, routing patches, or release claims.
Logging templates are synthetic and redacted. They do not include private logs, production execution logs, private repo paths, customer data, or Foundation-private data.
Refusal rules specify when agents must stop, hold, or escalate for human authority.
Run scripts/validate-public-boundary.sh, the plan-listed boundary/status rg scan, git diff --check, git status --short, and git remote -v.
This repo contains public-safe agentic engineering workflow patterns only. It must not contain active agent workforce claims, autonomous production agent claims, internal specialist prompts, private agent memory, sealed agent memory, live tool tokens, credentials, private repo paths, customer data, Foundation-private data, production execution logs, private runtime details, internal coordination layer internals, released software claims, or internal company product names.
Profile routing and proof-stack routing are planned only. Any links from Franzabner or franzabner-proof-stack require separate reviewed patches.
First artifact completed as scaffolded: execution-loops/codex-style-execution-loop.md covers synthetic agentic execution context, human authority rule, specialist role pattern, tool permission matrix summary, review gate lifecycle, synthetic task lifecycle, logging and refusal rules, human approval boundary, Mermaid diagram, validation questions, proof limits, and public/private/sealed checklist.