write-code

name	write-code
description	Edit source code for the current task. Delegates editing and lint/typecheck/self-repair to a subagent.
model	sonnet
user-invocable	false

Edit source code for the current task.

Procedure

When a hook blocks a sanctioned command, a fallback path is taken, or any unexpected outcome occurs, report via /supervisor-report (trigger conditions: rules/supervisor-reporting.md).

Read rules/ops.md before any destructive or system-state-changing command (including inside WCD-4 self-repair) — on-demand-only, never auto-injected; it owns the recovery-options-first decision path.

WCD-1. Read rules/core-principles.md, rules/coding.md (on-demand-only: not auto-injected, so this Read is the only way it arrives), and the target files identified from the plan.

WCD-2. CONFIRM_CODE gate — enumerate planned edits (one line per file: path + change intent). Then check via Bash: bash -c 'cd "$AGENTS_CONFIG_DIR" && bash "$AGENTS_CONFIG_DIR/bin/confirm-off" CONFIRM_CODE on'

• stdout OFF: proceed to step WCD-3.
• stdout ON or ERROR: present the planned edits via AskUserQuestion and wait for approval before continuing.

WCD-3. Run bash -c 'node "$AGENTS_CONFIG_DIR/bin/workflow/read-complexity-evaluation" --session "$SESSION_ID"'. If output is not NONE, use the stored level and signals directly (parse level=<v> and signals=<csv>), then derive the model: level === "high" ? "opus" : "sonnet". If NONE (fail-open): read skills/_shared/judge-task-complexity.md and evaluate directly → derive high/low from the routing rule, then map to model: high→opus, low→sonnet. Emit in Claude text output (NOT Bash echo):

Model selected: [opus|sonnet] (signals: [comma-separated triggered signal IDs, or "none"]) WCD-3a. Emit echo "<<WORKFLOW_MARK_STEP_write_code_in_progress>>" via Bash immediately before the WCD-4 subagent launch.

WCD-4. Launch subagent (Agent tool, mode: "default", model: <model derived from level in step WCD-3> — the model parameter receives "opus" or "sonnet", never "high" or "low") with a prompt containing:

• Target files and planned edit summary from step WCD-2.
• A-layer language essence block (see below).
• Directive: "Read rules/coding.md (the hub — on-demand-only, so it does not reach you otherwise) and rules/coding/<lang>.md for each language present, before the first Edit."
• Directive: "Read rules/shell-commands.md before the first Bash command — general-purpose dispatch does not inherit auto-injected rules."
• Directive: "Read rules/user-escalation.md before any system-state-changing command — general-purpose dispatch does not inherit auto-injected rules."
• Lint/typecheck recipe table (see below).
• Self-repair cap: 3 iterations; if still failing after 3, surface tool output verbatim.
• Lint-tool absence policy: when a tool is unavailable, skip that check AND emit <tool> not found — check skipped in the final summary. Never omit this notice.
• Scope-expansion policy: if editing reveals additional files not in the original list need changes, include them in the final summary with a reason. Do NOT prompt mid-edit; do NOT silently expand scope.
• Prohibitions: no diffs shown in the conversation; no mid-edit confirmation prompts.

WCD-5. Parse the subagent summary. Surface tool output on failure. Collect all check skipped notes and scope-expansion notes.

WCD-5a. Run skills/write-code/scripts/self-check-siblings.sh "<what changed>" "<why>" — CPR-E2C sibling-sweep reminder. WCD-5b. Run skills/write-code/scripts/detect-contract-pins.sh <edited-files> — flag edited files with no matching test; fold into WCD-6.

WCD-6. Present the final edited file list + skipped-check notes + scope-expansion notes to the user — gated by CONFIRM_CODE gate (post-action review): bash -c 'cd "$AGENTS_CONFIG_DIR" && bash "$AGENTS_CONFIG_DIR/bin/confirm-off" CONFIRM_CODE on'

• stdout OFF: skip this step; proceed (no user wait).
• stdout ON or ERROR: present the file list and notes.

A-layer language essence (complement of B-layer — zero overlap with rules/coding/*.md)

• Python: read rules/coding/python.md before the first Python Edit. All Python invariants (including modern type syntax) are owned by the B-layer.
• Node/JavaScript: read rules/coding/nodejs.md before the first JS Edit. All Node invariants (including module-system guidance) are owned by the B-layer.
• TypeScript (B-layer in rules/coding/nodejs.md; items below are writing-moment additions not in B-layer): discriminated union over loose interface; explicit return types on exported functions.
• PowerShell (no B-layer file): approved verbs; $ErrorActionPreference = 'Stop'; Set-StrictMode -Version Latest; full cmdlet names; $env:VAR for env vars.
• Bash (no B-layer file): set -euo pipefail; [[ ]] over [ ]; quote variable expansions; $(...) over backticks; ${var:?error} for required vars.
• JSON (no B-layer file): double-quoted keys/strings; no trailing commas; no comments; 2-space indent.
• YAML (no B-layer file): 2-space indent; no tabs; quote ambiguous scalars; block style for multi-line strings.

When a new standalone B-layer file is added for a language that currently has only A-layer content (PowerShell, Bash, JSON, YAML), that language's entry collapses to a bare read-directive. Test p in tests/feature-write-code-skill-static.sh enforces SSOT non-duplication for module-system guidance tokens at CI.

Lint/typecheck recipes

Language	Command
Python	uv run ruff check <file>
JavaScript	npx eslint <file>
TypeScript	npx eslint <file> + npx tsc --noEmit
PowerShell	pwsh -NoProfile -Command Invoke-ScriptAnalyzer -Path <file>
Bash	shellcheck <file>
JSON	node -e "JSON.parse(require('fs').readFileSync('<file>','utf8'))"
YAML	uv run python -c "import yaml,sys;yaml.safe_load(open(sys.argv[1]))" <file>

Each is best-effort: if the tool or config is absent, skip AND emit <tool> not found — check skipped.

Rules

• Mode-orthogonal: behavior is identical regardless of worktree mode. Do not show diffs in the conversation.
• Never edit test files — /write-tests owns them.
• Subagent self-repair cap: 3 iterations.
• Report observations via /supervisor-report (trigger conditions: rules/supervisor-reporting.md).

Completion

Emit echo "<<WORKFLOW_MARK_STEP_write_code_complete>>" via Bash once WCD-6 passes; skip it when the subagent failed or the WCD-6 review was rejected — fix the work and re-run WCD-4 first.