Folds Kelsey's observation that AI 'software factories' are a remix
of existing SDLC (CI/CD, compilers, gRPC, low-code) with added token
costs. Ties to MVI principle: does the AI add new capability, or just
burn tokens on work that commodity hardware already handles?
Folds antirez's observation that LLM export controls are futile
(weights leak, proxies bypass, enforcement requires dismantling
encryption) into the existing insider-threat model note. Argues
the real safeguard is agent-level sandboxing (OpenShell pattern),
not model distribution control — assume any model can run on any
hardware.
Folds the Adel + Chris Murphy talk (youtube.com/watch?v=SVWmuJx0hHM)
into loop_engineering.md. Covers the third inflection point
(ChatGPT → DeepSeek → OpenClaw autonomous agents), agent =
model + harness, OpenShell secure runtime with policy-based
egress and secrets management, AIQ multi-agent deep research
blueprint (intent router → orchestrator → specialists →
synthesizer at 50% lower cost), NemoClaw blueprints, Nemo Relay
observability with ATIF traces, and the Verified Skills pattern.
Maps each component to TAC's existing factory infrastructure.
Folds the plan-execute model split (68%+ cost savings) into
loop_engineering.md. Three-tier TAC model mapping: ZAI GLM-5.2
for frontier planning, Go/DeepSeek V4 Flash for execution,
Haiku/StepFun for fast/cheap tasks. Includes the Coinbase
pattern (open-source models for most work) and thinking-effort
tiering.
Folds Jeremy Howard's SDT-based talk (youtube.com/watch?v=SUZwYV5JYBM)
into loop_engineering.md. Covers autonomy/mastery/relatedness/purpose,
the dark flow trap in vibe coding, and the augmentation-vs-replacement
distinction. Four application points for agentic engineering loops.
Folds kunchenguid dotfiles architecture (Nix + Home Manager + symlinks
+ rebuild.sh) into loop_engineering.md as a cross-OS reference. Covers
the Windows port sketch (scoop/winget, Windows Terminal, Herdr native)
and the key insight: agent configs as versioned symlinks from one
source-of-truth agents.md file.
Reference: youtube.com/watch?v=5N-okeDdIuI, github.com/kunchenguid/dotfiles
Fold cmux and herdr agent-multiplexer patterns into TAC loop-engineering guidance. Treat the primary agent as orchestrator over panes/tabs via CLI + socket (no MCP required), pane as isolation boundary, and multiplexer state (working/idle/blocked) as the feedback signal. Keep TAC invariants: deploy stays signed-action-only via git-proxy:8099/deploy, peer agents remain insider-threat candidates with least privilege and receipts, multiplexer driving shells does not grant a raw-command deploy path.
Fold the three-video business, macro, and AI-learning synthesis into TAC loop-engineering guidance. Convert market stories, money-flow narratives, and AI demos into small receipt-bound operating loops with niche/distribution, money-flow, AI-amplifier, and seven-day experiment checks.
Capture the Security Fest talk in TAC loop-engineering guidance. Treat agents with file, tool, code, or deploy access as insider-threat candidates and require least privilege, no raw secrets, sandboxing, supply-chain checks, human approval for business-impacting actions, receipts, and promotion attestation.
Capture the AI-era role archetype discourse as delivery activity modes rather than permanent job titles. Add Prototyper, Grower, Sweeper, Reviewer/Taster, Architect/Primitive keeper, and Operator/SRE invariants while preserving the guardrail that role collapse does not make expertise obsolete.
Deepen the Codex product-work guidance with explicit artifact stages, production-readiness receipts, model-timing assumptions, and a deletion/simplification review criterion so cheap implementation does not create ungoverned prototype slop.
Capture the Codex product-work interview in TAC loop-engineering guidance. Preserve PRD/spec discipline, prototypes, human taste/systems review, receipt-gated background agents, and explicit ownership even when implementation gets cheaper.
Ground TAC loop-engineering guidance in external production references: NVIDIA Secure Agent Workspace, Infisical Agent Vault, Anthropic agent evals, and OWASP AI Agent Security. Lock signed-action-only deploy posture and research markers with pytest.
Capture the NanoClaw autonomous work-agent blueprint in the TAC loop-engineering meta-prompt. Require plans to distinguish shared agent factories from personal work agents, isolate each agent, separate credentials through vault/proxy injection, define access policies, and include management/revocation surfaces. Tests lock the new markers.
Capture the GLM 5.2 harness video in the loop-engineering meta-prompt. Require portable context, report/review phases, quality gates, and model/cost notes so TAC plans can run on lower-cost or open-source models without relying on frontier-model taste.
Capture the Sean's AI Stories loop-engineering video in the TAC meta-prompt. Require plans to name harness context, memory, guardrails, traces, evals, and promotion criteria, and lock those markers with pytest.
artale