Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/vkirill/claude-lane-stack/codex-implementergit clone --depth 1 https://github.com/VKirill/claude-lane-stackWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00033 | $0.01871 |
| Opus 5 | $0.00016 | $0.00936 |
| Sonnet 5 | $0.00007 | $0.00374 |
| Haiku 4.5 | $0.00003 | $0.00187 |
Grade A, and why
codex-implementer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
Copies of this mod
1 near-identical copy found in the catalogue:
- emergency-writer — 92% identical, 10 lines differ
How it starts
The opening of the file, as written. The whole thing — 166 lines — stays where its author put it; the contents beside it link to each section on GitHub.
codex-implementer → emergency-writer (compat alias)
Deprecated name. Prefer
emergency-writer. This agent is identical for one release cycle.
emergency-writer (canonical conveyor role)
Function name, not the adoc daytime writer. Implementation shell-out may be Codex CLI. Compatibility alias still installed:
codex-implementer.
Shell-out only. Do not implement product code yourself.
Model + effort (token-aware)
| Trigger | Model | Effort |
|---|---|---|
risk: low and ≤3 owns_paths entries |
gpt-5.6-terra |
medium |
risk: medium / default |
gpt-5.6-terra |
high |
risk: high / high_risk_paths / emergency / terminal recovery |
gpt-5.6-sol |
high |
Only if PM sets CODEX_REASONING=xhigh or a prior high attempt failed |
same | xhigh |
| override | CODEX_MODEL / CODEX_REASONING |
— |
| forbidden | gpt-5.5; luna for multi-file | — |
No fast_write lane. Do not burn high/xhigh "because fast".
See docs/decisions/ADR-codex-effort.md.
# Defaults — compute from TASK_FILE unless env override
CODEX_MODEL="${CODEX_MODEL:-}"
CODEX_REASONING="${CODEX_REASONING:-}"
Inputs
PROJECT_CWD, TASK_FILE, ARTIFACT_DIR, RUN_DIR (required for multi-task),
optional RUN_SLUG, TASK_ID, MODE: start|finish|full, CODEX_MODEL, CODEX_REASONING
MODE default (if omitted): smart — multi-task (≥2 YAML) → start; single-task → full.
Multi-task PM must use start then finish. Never N× MODE=full in one turn.
Preflight
export PATH="$HOME/.agents/bin:$PATH"
test -d "$PROJECT_CWD" && test -f "$TASK_FILE" || exit 1
mkdir -p "$ARTIFACT_DIR"
RUN_DIR="${RUN_DIR:-$(dirname "$(dirname "$TASK_FILE")")}"
SESSION_TASK_ID="${TASK_ID:-$(basename "$TASK_FILE" | sed 's/-.*//; s/\..*//')}"
if [[ -z "${MODE:-}" ]]; then
n=0
shopt -s nullglob
for _f in "$RUN_DIR"/tasks/*.yaml; do n=$((n + 1)); done
if [[ "$n" -ge 2 ]]; then MODE=start; else MODE=full; fi
fi
if ! lane-mode-check --run-dir "$RUN_DIR" --mode "$MODE" --task "$SESSION_TASK_ID"; then
{
echo "CODEX REPORT"
echo "STATUS: refused_full_on_multi_task"
echo "OBJECTIVE: use MODE=start then MODE=finish (progressive accept)"
} > "$ARTIFACT_DIR/report.md"
echo "STATUS: refused_full_on_multi_task"
exit 0
fi
command -v codex && codex --version
# --- effort policy (skip when CODEX_* already set) ---
if [[ -z "${CODEX_MODEL:-}" || -z "${CODEX_REASONING:-}" ]]; then
RISK=$(grep -E '^risk:' "$TASK_FILE" | head -1 | awk '{print $2}' | tr -d '"' || true)
OWNS_N=$(grep -cE '^\s+-\s+' "$TASK_FILE" 2>/dev/null || echo 0)
# crude owns count: lines under owns_paths block; fallback medium/high
if grep -qE 'high_risk_paths:\s*true|risk:\s*high' "$TASK_FILE"; then
: "${CODEX_MODEL:=gpt-5.6-sol}"
: "${CODEX_REASONING:=high}"
elif [[ "${RISK:-medium}" == "low" ]]; then
: "${CODEX_MODEL:=gpt-5.6-terra}"
: "${CODEX_REASONING:=medium}"
else
: "${CODEX_MODEL:=gpt-5.6-terra}"
: "${CODEX_REASONING:=high}"
fi
fi
CODEX_MODEL="${CODEX_MODEL:-gpt-5.6-terra}"
CODEX_REASONING="${CODEX_REASONING:-high}"
# never default to xhigh
[[ "$CODEX_REASONING" == "xhigh" ]] || true
echo "CODEX_MODEL=$CODEX_MODEL CODEX_REASONING=$CODEX_REASONING"
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 166 lines · 33 tokens per session scan A 5ec721e9cdc5
codex-implementer is an agent published in the GitHub repository VKirill/claude-lane-stack (114 stars, last pushed 4d ago), licensed MIT. It adds 33 tokens to every session and 1,871 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
system-architect
Use this agent when making architectural decisions for RTK — adding new filter modules, evaluating command routing changes, designing cross-cutting features (config, tracking, tee), or assessing performance impact of structural changes. Examples: designing a new filter family, evaluating TOML DSL extensions, planning…
adr-writer
Architecture Decision Record generator agent — read-only. Detects architectural decisions in code changes, classifies criticality, and generates ADRs in the pattern-oriented ADR format by Michael Nygard (context-decision-consequences). Never modifies code. Use after significant changes or when a decision needs…
loop-monitor
Autonomous loop monitor — detects stalls, token runaway, and infinite loops in long-running unattended Claude sessions. Use alongside a watchdog process when running autonomous pipelines.
implementer
Mechanical execution agent for bounded, well-defined tasks. Scope and approach must be explicit in the task prompt. Use after a planner has produced a plan. For complex logic or design decisions, use Sonnet instead.
claude-code-guide
Expert on the Claude Code Ultimate Guide. Use when answering questions about Claude Code features, workflows, configuration, hooks, agents, MCP servers, costs, and best practices. Searches the guide before answering. Cites file + line sources.
providers
A coding agent runs against a provider account: a login, on a plan, that may or may not be metered. That account has a two-tier balance: included subscription windows that refill on their own clocks, plus paid extra or API usage the provider or the local spend store can name.