Codex System — Planning, Design & Complex Implementation

Codex CLI handles planning, design, and complex code implementation.

Preflight (SSOT): Update CLIs before each session — claude update && npm install -g @openai/codex@latest. Releases drift frequently (model names, flags, sandbox semantics). Other skills reference this line instead of repeating it. Delegation policy (when to delegate): .claude/rules/codex-delegation.md

Two Roles of Codex

1. Planning & Design

• Architecture design, module composition
• Implementation plan creation (step breakdown, dependency ordering)
• Trade-off evaluation, technology selection
• Code review (quality and correctness analysis)

2. Complex Implementation

• Complex algorithms, optimization
• Debugging with unknown root causes
• Advanced refactoring
• Multi-step implementation tasks

When to Delegate

Delegation policy — when to consult, when NOT to, and trigger criteria — lives in .claude/rules/codex-delegation.md (SSOT). This skill covers how to consult. The other half of "how" is how to verify what came back: root AGENTS.md → Guardrails (Completion Verification) is mandatory for every write-access call, and Verify Before Trusting below is its executable form. A delegated CLI is never trusted on its self-report.

How to Consult

Invoke Codex through the wrapper — .claude/skills/_shared/codex_consult.py — instead of calling codex exec directly. codex exec itself waits for stdin EOF and hangs indefinitely when stdin is left open (e.g. background shells); the wrapper always runs it with stdin closed, so callers never need < /dev/null. It also passes the prompt as a single argv element (no shell, so nested quotes in the prompt body never break it), captures stdout/stderr to timestamped files under .claude/logs/codex/, and reports one JSON result instead of silently discarding stderr.

python3 .claude/skills/_shared/codex_consult.py (--prompt-file PATH | --prompt-stdin) [--label L] [--caller AGENT] [--sandbox {read-only,workspace-write,danger-full-access}] [--model M] [--timeout N] [--cwd DIR] [--project-root DIR] [--skip-git-repo-check] [--config KEY=VALUE]

Consulting a peer CLI instead of Codex. Claude Code and Antigravity go through .claude/skills/_shared/cli_consult.py --cli {claude,antigravity} --prompt-file PATH, unrestricted unless --read-only — which Antigravity refuses, because its headless mode auto-approves every tool call and the restriction cannot be enforced from the caller (--resume SESSION is Claude-only; --cli-arg forwards a native flag; default timeout 900 s; same four exit codes as below). Never shell out to a CLI directly — the wrapper-only rule and the per-callee permission mapping live in root AGENTS.md. Codex's sandbox and --config semantics stay here because they are Codex-specific; everything cross-CLI is in that rule file.

• Write the prompt body (Objective / Constraints / Relevant files / Acceptance checks / Output format) to a file and pass it via --prompt-file; use --prompt-stdin to pipe a short prompt instead. Any path works — the ad-hoc snippets below use mktemp, while skills write to .claude/logs/codex/prompt-{label}.md so the prompt sits next to the response the wrapper writes for it, which is what makes a disappointing answer diagnosable afterwards.
• --sandbox defaults to danger-full-access, matching .codex/config.toml. Pass --sandbox read-only explicitly for planning and review calls — see Sandbox Modes below.
• --model defaults to $CODEX_MODEL, else gpt-5.6-sol. --label is a [a-z0-9-]+ slug used in the log filenames (default consult). --timeout defaults to 600 seconds. --skip-git-repo-check covers the non-Git working directory case — see references/troubleshooting.md.
• --config KEY=VALUE (repeatable) forwards a Codex config override, e.g. --config model_reasoning_effort=low for a cheap question. Keys naming a sandbox or approval setting are refused: --sandbox must stay the single visible statement of what Codex is allowed to touch.
• The wrapper prints exactly one JSON object: {ok, exit_code, model, sandbox, write_access, timed_out, duration_sec, response_file, stderr_file, response_chars, response_head, error}. response_head is only a ~400-char preview — read the file at response_file for the full response, and stderr_file (non-null whenever Codex wrote to stderr) when diagnosing a failure.
• Exit codes: 0 succeeded · 1 bad args or unreadable prompt file · 2 codex not on PATH · 3 codex exited non-zero or timed out.

Subagent Pattern (Recommended)

Task tool parameters:
- subagent_type: "general-purpose-opus"
- run_in_background: true (optional)
- prompt: |
    Consult Codex about: {topic}

    Write the prompt body below to a file, then run the wrapper against it:

    Objective: {single-sentence objective}
    Constraints:
    - {constraint 1}
    Relevant files:
    - {file paths}
    Acceptance checks:
    - {commands}
    Output format:
    ## Analysis
    ## Recommendation
    ## Implementation Plan
    ## Risks
    ## Next Steps

    python3 .claude/skills/_shared/codex_consult.py --prompt-file {prompt_path} --label {short-slug} --sandbox read-only

    Parse the JSON result. ok: true means only that codex exec exited 0 — it is
    not a completion report. Read response_file for the full analysis, and judge
    it yourself: state which claims you verified and which you could not.
    Return CONCISE summary (key recommendation + rationale + what is unverified).

For a write-access subagent call, add the Verify Before Trusting steps to the delegated prompt as well, and require the subagent to return the verify.sh and verify_delegation.py verdicts. A subagent that summarises Codex's self-report without them has not verified anything, and its summary must not be treated as a completion.

Direct Call (short questions, responses up to ~50 lines)

echo "Objective: {brief question}" | python3 .claude/skills/_shared/codex_consult.py --prompt-stdin --label quick-question --sandbox read-only

Having Codex Implement Code

prompt_file="$(mktemp)"
cat > "${prompt_file}" << 'EOF'
Objective: Implement {detailed implementation task}
Constraints:
- Follow existing project conventions
- Keep diffs minimal
Relevant files:
- {file paths}
Acceptance checks:
- {commands}
Output format:
## Changes Made
## Validation
## Remaining Risks
EOF
python3 .claude/skills/_shared/codex_consult.py --prompt-file "${prompt_file}" --label implement --sandbox danger-full-access

The call is not finished here. Continue with the next section.

Verify Before Trusting

Mandatory after every workspace-write / danger-full-access call (Codex or a peer CLI), per root AGENTS.md → Guardrails. ok: true from the wrapper means only that codex exec exited 0; it says nothing about whether the change is correct, complete, or honest.

1. Run the acceptance checks from your own prompt, plus the project gates:

bash .claude/skills/_shared/verify.sh

Exit 0 = overall: "pass". Exit 2 = a gate failed, or no gate ran at all (overall: "no_gates") — a delegated code change must never be accepted with zero checks executed. Exit 1 bad arguments, 3 the log file could not be written. Read log_file for the full output.

2. Collect the Guardrail evidence from the diff, naming the scope the prompt actually authorised:

python3 .claude/skills/_shared/verify_delegation.py --base HEAD \
  --expect-files {file the task was supposed to change} \
  --forbid-outside {directory the task was scoped to}

It reports deletions, placeholders, weakened_tests, out_of_scope_files, missing_expected_files, scope_empty, and the captured diff at diff_file. Exit 0 = nothing actionable and no violated expectation — deletions alone land here, reported but not actionable on their own, and exit 0 is still not an accept. Exit 2 = an actionable finding (placeholders, weakened_tests) or a violated expectation (out_of_scope_files, missing_expected_files, scope_empty). Use --base <pre-delegation ref> when Codex committed its work; the default HEAD covers the usual uncommitted case.

3. Read the diff and decide. verdict is always needs-review and there is no verdict that means "accepted" — deliberately. The pattern list is heuristic (a legitimate test deletion exists, and a TODO in a docstring is not a stub), and only you know what the prompt authorised. Reject the completion when the diff shows any of:

• tests deleted, skipped (@pytest.mark.skip), or weakened (assertions removed or loosened) to make the suite pass;
• exceptions silently swallowed (except: pass or equivalent) to hide failures;
• hard-coded return values substituted for real logic — the one Guardrail item no script screens for, so it is listed under not_automated and only your read of the diff catches it;
• stub or placeholder completions where real logic was requested;
• files changed that the task never mentioned, or unapproved deletions.

4. On failure, follow the re-delegate-once protocol (cli-execution.md (c)): report the specific failures with evidence, re-delegate once with the original prompt plus the failure context appended, and if the second attempt also fails verification, halt and require explicit user approval before proceeding. Never patch over a failed delegation silently.

Sandbox Modes

Mode	Sandbox	Use Case
Analysis	read-only (explicit opt-in)	Design review, debugging, trade-off analysis
Implementation	danger-full-access (default)	Implementation, fixes, refactoring

The wrapper's own default is danger-full-access, the same value .codex/config.toml sets, so an implementation call needs no flag and an analysis call must pass --sandbox read-only explicitly. The wrapper used to default to read-only — deliberately stricter than a bare codex exec — which meant the access Codex had depended on whether it was reached through the wrapper or directly, and neither call site said so. Aligning the two removes that divergence; what has not changed is that the wrapper always sends --sandbox explicitly, so the granted access is readable in the command rather than inherited from a config file, and --config keys naming a sandbox or approval setting stay refused.

Because the default is unrestricted, every call is bracketed by an edit snapshot and the JSON result carries an edits object naming the files Codex created, changed, or deleted, plus caller and label. Pass --caller <your agent name> so .claude/logs/cli-tools.jsonl records which subagent asked for a change, not only which CLI made it.

Task Templates

Implementation Planning

prompt_file="$(mktemp)"
cat > "${prompt_file}" << 'EOF'
Create an implementation plan for: {feature}

Context: {relevant architecture/code}

Provide:
1. Step-by-step plan with dependencies
2. Files to create/modify
3. Key design decisions
4. Risks and mitigations
EOF
python3 .claude/skills/_shared/codex_consult.py --prompt-file "${prompt_file}" --label plan --sandbox read-only

Design Review

prompt_file="$(mktemp)"
cat > "${prompt_file}" << 'EOF'
Review this design approach for: {feature}

Context: {relevant code or architecture}

Evaluate:
1. Is this approach sound?
2. Alternative approaches?
3. Potential issues?
4. Recommendations?
EOF
python3 .claude/skills/_shared/codex_consult.py --prompt-file "${prompt_file}" --label design-review --sandbox read-only

Debug Analysis

prompt_file="$(mktemp)"
cat > "${prompt_file}" << 'EOF'
Debug this issue:

Error: {error message}
Code: {relevant code}
Context: {what was happening}

Analyze root cause and suggest fixes.
EOF
python3 .claude/skills/_shared/codex_consult.py --prompt-file "${prompt_file}" --label debug --sandbox read-only

Language Protocol

Ask Codex in English and receive English back. The user-facing report follows CLAUDE.md ## Language Protocol for language and .claude/rules/language.md ## Response Style for how that reply reads.

Codex Plugin Commands (codex-plugin-cc)

When the openai/codex-plugin-cc plugin is installed, these slash commands are available:

Plugin source: https://github.com/openai/codex-plugin-cc

Availability precondition: run /codex:setup first. Nothing in this repository verifies the plugin is installed, so if the commands are absent, use codex_consult.py instead of assuming a route exists. Audit-trail caveat: plugin commands run Codex outside the wrapper, so they produce no .claude/logs/codex/ response or stderr capture and no .claude/logs/cli-tools.jsonl entry (log-cli-tools.py keys on the wrapper filenames). Prefer the wrapper wherever both work, and note the gap when you use a plugin route for work that needs a record.

Code Review

/codex:review                    # Review current uncommitted changes
/codex:review --base main        # Review branch diff against main
/codex:review --background       # Run review in background
/codex:review --wait             # Synchronous: block until review finishes

Adversarial Review

/codex:adversarial-review                           # Challenge design decisions
/codex:adversarial-review --base main               # Branch-level adversarial review
/codex:adversarial-review --background look for race conditions

Task Delegation (Rescue)

/codex:rescue investigate why the tests started failing
/codex:rescue fix the failing test with the smallest safe patch
/codex:rescue --resume apply the top fix from the last run
/codex:rescue --model gpt-5.5-mini --effort medium investigate flaky test
/codex:rescue --background investigate the regression

Job Management

/codex:status                    # Check progress of background jobs
/codex:result                    # Show finished job output
/codex:cancel                    # Cancel active background job

Setup

/codex:setup                     # Check if Codex is installed and authenticated
/codex:setup --enable-review-gate   # Enable auto-review gate (use with caution)
/codex:setup --disable-review-gate  # Disable review gate

When to Use Plugin vs Direct CLI

Scenario	Use
Pre-ship code review	/codex:review
Challenge design	/codex:adversarial-review
Delegate investigation/fix	/codex:rescue
Background work + tracking	Plugin --background
Ad-hoc design question	codex_consult.py (direct)
Unrestricted implementation	codex_consult.py (the default) + Verify Before Trusting
Subagent delegation	codex_consult.py via general-purpose-opus
Consulting Claude Code or Antigravity	cli_consult.py --cli {claude,antigravity}

Plugin routes (the first four rows) leave no wrapper log, no cli-tools.jsonl entry, and no record of which files they changed; the codex_consult.py routes leave all three. Whichever route made the change, an unrestricted run is verified the same way.

Why Codex?

• Deep reasoning: Complex analysis and problem-solving
• Planning expertise: Architecture and implementation strategies
• Code mastery: Complex algorithms, optimization, debugging

References

Detailed templates and patterns in references/:

• agent-prompts.md — Prompt templates for specialized review agents (Architect, etc.)
• code-review-task.md — Prompt template for delegating code review to Codex
• delegation-patterns.md — Delegation decision flowchart and detailed patterns
• refactoring-task.md — Prompt template for delegating refactoring to Codex
• troubleshooting.md — Codex CLI troubleshooting (installation, auth, common errors)

Also .claude/docs/CODEX_HANDOFF_PLAYBOOK.md — the handoff templates .claude/rules/codex-delegation.md points at, kept there because they are shared with non-Codex handoffs.