// Use the Gemini CLI as a breadth-exploration co-pilot for paper-from-zero. Primary role: candidate direction expansion, adjacent work scanning, keyword cluster generation, and alternative framing proposals. Also supports generic code collaboration as fallback.
| name | collaborating-with-gemini |
| description | Use the Gemini CLI as a breadth-exploration co-pilot for paper-from-zero. Primary role: candidate direction expansion, adjacent work scanning, keyword cluster generation, and alternative framing proposals. Also supports generic code collaboration as fallback. |
| metadata | {"short-description":"Breadth co-pilot (Gemini) for paper-from-zero"} |
Use Gemini CLI as a breadth-exploration collaborator for paper-from-zero, or as a generic code collaborator in fallback mode.
This skill provides a lightweight bridge script that returns structured JSON
and supports multi-turn sessions via SESSION_ID.
Shared rules: This skill inherits all shared rules from
../../_shared/collaboration-base.md(core rules, candidates-only constraint, fallback code-collaboration mode, safety guardrails, and collaboration state capsule format). Only Gemini-specific additions are listed below.
ref.bib entries or verified citations.--sandbox to run Gemini in sandbox mode.GEMINI_CLI_HOME to a writable repo-local directory
(<repo>/.gemini_cli_home/) to avoid Gemini CLI hanging in some runners.gemini is already logged in, reuse that login with
--use-user-home.GEMINI_API_KEY set but want to use OAuth creds, add
--prefer-oauth to unset API-key env vars for this run.If your prompt contains Markdown backticks, use a heredoc; see references/shell-quoting.md.
PROMPT="$(cat <<'EOF'
Review src/auth.py around login() and propose fixes.
OUTPUT: Unified Diff Patch ONLY.
EOF
)"
python3 .codex/skills/collaborating-with-gemini/scripts/gemini_bridge.py --cd "." --PROMPT "$PROMPT"
Output: JSON with success, SESSION_ID, agent_messages, and optional error / all_messages.
# Start a session
PROMPT="$(cat <<'EOF'
Analyze the bug in foo(). Keep it short.
EOF
)"
python3 .codex/skills/collaborating-with-gemini/scripts/gemini_bridge.py --cd "." --PROMPT "$PROMPT"
# Continue the same session
PROMPT="$(cat <<'EOF'
Now propose a minimal fix as Unified Diff Patch ONLY.
EOF
)"
python3 .codex/skills/collaborating-with-gemini/scripts/gemini_bridge.py --cd "." --SESSION_ID "<SESSION_ID>" --PROMPT "$PROMPT"
Use assets/prompt-template.md as a starter when crafting --PROMPT.
Paper-mode templates (top of the file):
literature-map — expand candidate papers and adjacent research areasinnovation-candidates — generate 3-5 candidate contribution framingsevidence-matrix-gap-check — audit evidence coverage gapsEach template specifies its input sources, output artifact mappings, and constraints.
--sandbox: Run Gemini in sandbox mode.--model <name>: Override the default Gemini model.--return-all-messages: Include all raw messages (tool calls, traces) in output JSON.--use-user-home: Reuse your normal home (and existing ~/.gemini login) for this run.--prefer-oauth: Unset GEMINI_API_KEY/GOOGLE_API_KEY so OAuth creds can be used (if present in the selected home).--seed-user-auth: Copy OAuth/settings files from your ~/.gemini into <repo>/.gemini_cli_home/.gemini/ (avoids --use-user-home hangs).--fake-responses <path.jsonl>: Offline deterministic run using Gemini CLI fake responses (for smoke tests).check-collaborators skillpython3 .codex/skills/collaborating-with-gemini/scripts/gemini_bridge.py --help../../_shared/collaboration-base.md (shared rules)assets/prompt-template.md (prompt patterns)references/shell-quoting.md (shell quoting/backticks)Writes ML/AI review and survey papers for arXiv using the IEEEtran LaTeX template with verified BibTeX citations. Triggers when the deliverable is a review paper, literature survey, or when an existing LaTeX project needs citation validation or repair.
Verify that Gemini CLI and Claude Code CLI are installed, authenticated, and API-reachable before starting collaboration workflows.
Use the Claude Code CLI as a depth-analysis co-pilot for paper-from-zero. Primary role: claim tree compression, logical hole detection, evidence sufficiency audit, and review/empirical routing judgment. Also supports generic code collaboration as fallback.
Draft IEEE-style empirical ML/AI papers from a structured research contract. Builds experiment plans, section skeletons, placeholder-safe results, and a near-submission draft without fabricating evidence.
Post-process LaTeX project prose to improve readability through varied sentence and paragraph lengths. Removes filler phrases and unnecessary transitions while preserving all citations and semantic meaning.
Route a fixed research topic into a rigorous paper-generation workflow. Handles active literature search, innovation framing, contribution/evidence planning, and routes to either the review writer or empirical writer skill.