Ln 403 Task Rework

> Paths: File paths (references/, ../ln-*) are relative to this skill directory.

Task Rework Executor

Type: L3 Worker

Executes rework for a single task marked To Rework and hands it back for review.

Purpose & Scope

• Load full task, reviewer comments, and parent Story; understand requested changes.
• Apply fixes per feedback, keep KISS/YAGNI, and align with guides/Technical Approach.
• Update only this task: To Rework -> In Progress -> To Review; no other tasks touched.

Hex-line acceleration (if available): Use outline(file_path) before reading large code files. Use read_file() for discovery and read_file(edit_ready=true, verbosity="full") before any edit that needs revision and checksums. After edits: edit_file(base_revision=rev) → verify(checksums). Use changes() to show what was fixed.

Inputs

Use read_file() and edit_file() as the primary path for code/config/script/test files during rework. Keep read_file() discovery-first by default; request edit_ready=true, verbosity="full" only when you are about to reuse its revision/checksum protocol. Built-in Read/Edit are fallback only when hex-line is unavailable.

| Input | Required | Source | Description | |-------|----------|--------|-------------| | taskId | Yes | args, parent Story, kanban, user | Task to rework |

Resolution: Task Resolution Chain. Status filter: To Rework

Task Storage Mode

MANDATORY READ: Load references/environment_state_contract.md, references/storage_mode_detection.md, references/tracker_provider_contract.md, and references/input_resolution_pattern.md

Extract: task_provider = Task Management → Provider (linear | github | file). Operations stay provider-agnostic in this skill — see references/tracker_provider_contract.md for the canonical operation set and provider_*.md for transport binding.

Tracker operations used by this skill: getTask, getStory, listComments, updateStatus (To Rework → In Progress → To Review), addComment. Transport per provider lives in references/provider_file.md, references/provider_github.md, references/provider_linear.md.

MANDATORY READ: Load references/mcp_tool_preferences.md — ALWAYS use hex-line MCP for code files when available. No fallback to standard Read/Edit unless hex-line is down.

Workflow (concise)

1) Resolve taskId: Run Task Resolution Chain per guide (status filter: [To Rework]). 2) Load task: Read task and review notes via the configured tracker provider (getTask, listComments); fetch parent Story via getStory. 2b) Goal gate: MANDATORY READ: Load references/goal_articulation_gate.md — State REAL GOAL of this rework (what was actually broken, not "apply feedback"). Combine with 5 Whys (references/problem_solving.md) to ensure root cause is articulated alongside the rework goal. NOT THE GOAL: a superficial patch that addresses the symptom, not the cause. 3) Plan fixes: Map each comment to an action; confirm no new scope added. 4) Implement: MANDATORY READ: Load references/code_efficiency_criterion.md — Follow task plan/checkboxes; address config/hardcoded issues; update docs/tests noted in Affected Components and Existing Code Impact. Before handoff, verify 3 efficiency self-checks. 5) Quality: Run typecheck/lint (or project equivalents); ensure fixes reflect guides/manuals/ADRs/research. 6) Root Cause Analysis: Ask "Why did the agent produce incorrect code?" Classify: missing context | wrong pattern | unclear AC | gap in docs/templates. If doc/template gap found → update the relevant file (guide, template, CLAUDE.md) to prevent recurrence. 7) Handoff: Set task to To Review via the configured tracker provider (updateStatus); move it in kanban; add summary comment referencing resolved feedback + root cause classification.

Critical Rules

• Single-task only; never bulk update.
• Do not mark Done; only To Review (the reviewer decides Done).
• Keep language (EN/RU) consistent with task.
• No new tests/tasks created here; only update existing tests if impacted.
• Do NOT commit. Leave all changes uncommitted — the reviewer reviews and commits.

Runtime Summary Artifact

MANDATORY READ: Load references/coordinator_summary_contract.md, references/worker_runtime_contract.md, references/task_worker_runtime_contract.md

Shared contract:

• emit summary_kind=task-status
• standalone mode omits runId and summaryArtifactPath
• managed mode passes both runId and exact summaryArtifactPath before the worker writes its validated summary

Monitor (2.1.98+): When verification commands expected >30s, use Monitor. Fallback: Bash(run_in_background=true).

Definition of Done

• [ ] Task and review feedback fully read; actions mapped.
• [ ] Fixes applied; docs/tests updated as required.
• [ ] Quality checks passed (typecheck/lint or project standards).
• [ ] Root cause classified (missing context | wrong pattern | unclear AC | doc gap); doc/template updated if gap found.
• [ ] Status set to To Review; kanban updated; summary comment with fixed items + root cause.
• [ ] Runtime summary artifact written to the shared task-status location.

Reference Files

• Environment state: references/environment_state_contract.md
• Storage mode operations: references/storage_mode_detection.md
• [MANDATORY] Problem-solving approach: references/problem_solving.md
• Kanban format: docs/tasks/kanban_board.md

---

Version: 3.0.0 Last Updated: 2025-12-23

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: levnikolaevich
• Source: levnikolaevich/claude-code-skills
• License: MIT
• Homepage: https://levnikolaevich.github.io/claude-code-skills/

Install and usage instructions live in the source repository linked above.