gemini-phase-loop

Bridges Gemini CLI to the repo-local phase-loop runner for status, resume, and bounded run flows.

2|1|Updated Jun 25, 2026
One-click install
npx skills add https://github.com/Consiliency/agent-harness --skill gemini-phase-loop-consiliency
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: gemini-phase-loop
Source: https://github.com/Consiliency/agent-harness/tree/main/skills-src/gemini/gemini-phase-loop
Command: npx skills add https://github.com/Consiliency/agent-harness --skill gemini-phase-loop-consiliency

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Operators using Gemini CLI need a consistent way to inspect and drive the repo-local phase-loop orchestration runner without reimplementing phase selection, reconciliation, or event handling, and without misreading shared loop state. ## Core Features & Use Cases - Command Mapping: Maps Gemini-facing commands (handoff, status, monitor, run, dry-run, sync-skills) to the shared phase-loop / codex-phase-loop CLI surface. - Shared State Inspection: Reads .phase-loop/state.json, events.jsonl, handoff files, and per-run heartbeat/terminal-summary artifacts to report automation and verification status. - Drift and Blocker Handling: Detects installed-skill drift, frozen blockers such as missing_secret or branch_sync_conflict, and routes spec-delta closeout decisions through the runner evidence contract. - Use Case: A Gemini operator resumes an unattended phase-loop run after another harness drove the loop, starting from handoff or monitor --once --json to avoid acting on stale TUI state. ## Quick Start Ask the assistant to check the current phase-loop status and handoff state using the gemini-phase-loop bridge before resuming a bounded run with --max-phases 1.

Frequently Asked Questions about gemini-phase-loop

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a bounded phase-loop run from Gemini CLI?▼

Use the mapped command `phase-loop run --max-phases <N> --closeout-mode manual`, preferring `--max-phases 1` unless you name another bound. The bridge does not reimplement the runner; it delegates to the repo-local phase-loop CLI.

How do I check phase-loop status and monitor state?▼

Use `phase-loop status` or `phase-loop status --json` for exact machine state, and `phase-loop monitor --once --json` for a single monitoring snapshot. These read the shared `.phase-loop/` state files rather than harness-private data.

What should I do if the Gemini skill bridge reports drift?▼

If the advisory bridge record reports `missing_root`, `missing_skill`, or `drifted`, repair it with `codex-phase-loop sync-skills --apply` before trusting manual Gemini reentry. Installed skills are a manual reentry surface, not a requirement for the runner to launch.

Why does a pipeline-mode dispatch fail with branch_sync_conflict?▼

With branch governance on, dispatch switches to a pipeline branch cut from `origin/main`; if the roadmap is unpushed or ahead of the base, the runtime guard blocks with `branch_sync_conflict`. Push the roadmap to the base or pass `--allow-branchgov`.

Does a missing Gemini trusted-workspace bypass mean the skill failed?▼

No. Headless disposable Gemini runs may need a trust bypass such as `--skip-trust`, but missing trust is treated as launch posture metadata, not as installed-skill failure.