task_tool_finalize

An orchestration command that delegates unfinished implementation work to specialised subagents in sequence. It is intended to coordinate the work rather than edit the project directly.

In plain words
What is it for?
Use it after a task reaches its cycle limit, receives a revision request, or leaves lint, TypeScript, build, or validation problems that need further work.
Why use it?
It provides a recovery path when an earlier implementation task still has errors, warnings, or unfinished review changes.

Command

Install

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.

agentmods
npx agentmods add commands/codeoutin/claude-code-agency/task_tool_finalize
Clone the repo
git clone --depth 1 https://github.com/codeoutin/claude-code-agency
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,282 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.01282
Opus 5 $0.00000 $0.00641
Sonnet 5 $0.00000 $0.00256
Haiku 4.5 $0.00000 $0.00128

Measured 2d ago against content hash 73aa082df2a0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

task_tool_finalize 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.

commands/task_tool_finalize.md · 155 lines

How it starts

The opening of the file, as written. The whole thing — 155 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Tool Finalize Command - Complete Remaining Implementation Work

YOUR ROLE: ORCHESTRATOR ONLY

YOU ARE THE ORCHESTRATOR. You must NOT write any code, read files for implementation, or make any edits yourself. Your ONLY job is to call the Task tool to spawn specialized subagents in sequence. If you find yourself writing code or editing files, STOP — you are doing it wrong. Delegate everything.

When to Use This Command

Use /task_tool_finalize when:

  • /task_complete reached the 5-cycle limit but didn't achieve 100% completion
  • Quality reviewer marked implementation as NEEDS_REVISION in final cycle
  • Lint or TypeScript errors remain after initial implementation
  • Build passes but still has warnings or validation bypasses

SETUP

$ARGUMENTS contains the path to an existing task directory from a previous /task_complete run.

Set TASK_DIRECTORY = $ARGUMENTS. Verify it contains: plan.md, context.md, validation-tracking.json.


STEP 1 — CALL Task tool: tc-context-gatherer

IMMEDIATELY call the Task tool with these parameters.

  • subagent_type: tc-context-gatherer
  • description: "Analyze remaining work"
  • prompt:
    Task directory: {TASK_DIRECTORY}
    Mode: FINISH_MODE
    
    Read all existing files in {TASK_DIRECTORY}/ — implementation reports, review reports,
    validation-tracking.json. Identify all remaining lint errors, TypeScript errors, and
    build issues with exact file locations. Save to {TASK_DIRECTORY}/finish-context.md with:
    - Current completion percentage
    - Specific errors remaining (file:line format)
    - Priority order for fixes
    

BLOCKING GATE: Do NOT proceed to Step 2 until the Task tool call returns.


STEP 2–3 — Error-Fixing Loop (Maximum 20 cycles)

Set cycle = 1. Target: zero errors, clean build.

STEP 2 — CALL Task tool: tc-implementation-agent

IMMEDIATELY call the Task tool with these parameters.

  • subagent_type: tc-implementation-agent
  • description: "Fix errors (finalize cycle {cycle})"
  • prompt:
    Task directory: {TASK_DIRECTORY}
    Mode: FINISH_MODE
    Cycle: {cycle}
    
    Read {TASK_DIRECTORY}/finish-context.md for the list of remaining errors.
    Fix ONLY lint errors, TypeScript errors, and build issues. Do NOT add new features.
    Work through errors systematically. Run npm run lint and npm run build after each fix.
    Save output to {TASK_DIRECTORY}/finish-lint-{cycle}.txt and {TASK_DIRECTORY}/finish-build-{cycle}.txt.
    Update {TASK_DIRECTORY}/validation-tracking.json with current error counts.
    Save report to {TASK_DIRECTORY}/finish-implementation-{cycle}.md.
    
    Target state: lint_errors = 0, typescript_errors = 0, build_success = true
    

Read the full file on GitHub · 155 lines

Changes

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.

  1. 2d ago First seen · 155 lines · 0 tokens per session scan A 73aa082df2a0

Subscribe to this mod's changes

task_tool_finalize is a command published in the GitHub repository codeoutin/claude-code-agency (41 stars, last pushed 6mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,282 tokens. 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.