graph-deploy

name	graph-deploy
description	Execute the DEPLOY phase via the `agf` CLI — release, DORA metrics, provider choice, cost proof; strictest gate (harness ≥ 70). Zero MCP
triggers	["graph-deploy"]
version	2.0.0
author	auto-generated
date	"2026-06-16T00:00:00.000Z"
category	DEPLOY
phase	DEPLOY
tokens	~623
phases	["HANDOFF","LISTENING"]

graph-deploy

Release, DORA, provider choice, cost proof. Drive via the agf CLI — zero MCP. Load context with agf context <id> before changing anything.

When to Use

• HANDOFF approved
• Preparing a release (strictest gate: harness ≥ 70)

Mandatory Flow

agf gate deploy → agf forecast → agf metrics --simulate

Steps (DEPLOY phase)

Command	Does
agf provider use <id>	choose provider (openrouter/copilot/ollama)
agf deliver "<request>" --live	autonomous end-to-end delivery
agf forecast	DORA (deploy freq, lead time, CFR, MTTR)
agf metrics --simulate	re-price real bill under all models (cost proof)
agf gate deploy	DEPLOY gate (release_check + harness ≥ 70)

Workflow

1. Gate — agf gate deploy (release_check + harness ≥ 70)
2. DORA — agf forecast
3. Cost proof — agf metrics --simulate
4. Provider — agf provider use <id> (if switching for release)
5. Deploy — run the project deploy pipeline
6. Post-deploy — verify health checks, logs, metrics
7. Baseline — agf snapshot create (post-deploy state)
8. Transition — agf phase LISTENING

Exit

• Release validated
• Harness ≥ 70 (B)
• DORA baseline saved
• Cost simulation reviewed

Output Format

Phase: DEPLOY → LISTENING
Release: validated  Harness: X (≥70)
DORA: deploy Y/day, lead P85 Zh, CFR W%, MTTR Vh
Cost: re-priced under N models
Snapshot: pre + post
Status: deployed, monitoring in LISTENING

Loop link → LISTENING (graph-listening): agf learning stats. Spiral: agf savings → agf learning → next.

Anti-Patterns

• Don't deploy with harness < 70 — strictest release gate
• Don't ignore cost — use --simulate for proof
• Don't deploy without a pre-deploy snapshot (rollback)
• Don't skip post-deploy validation

Related Skills

• $graph-handoff — agf skill show graph-handoff
• $graph-listening — agf skill show graph-listening

Codex Notes

• In Codex Plan Mode, plan only — don't mutate files.
• During implementation, follow AGENTS.md and use apply_patch for manual edits.

Token Economy

Economia de tokens. Os levers compartilhados por todas as skills — --select, agf retrieve-command, agf exec chain, reuso antes de criação — vivem em _shared.md → Token Economy. Fonte única: um parágrafo repetido em trinta arquivos é o trigésimo primeiro que envelhece sozinho.