implementation-agent

A workspace coding agent that implements work described by a GitHub issue and records progress in the issue’s comments.

In plain words
What is it for?
Finding the issue linked to a branch, implementing the next piece of work, updating progress, responding to review feedback, and merging approved changes.
Why use it?
It lets repeated runs continue from saved progress instead of starting over, while handling the usual branch, commit, push, and pull-request steps.

Skill for Claude CodeCodex

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 skills/band-app/band/implementation-agent
Any agent
npx skills add band-app/band --skill implementation-agent
Clone the repo
git clone --depth 1 https://github.com/band-app/band

Made for: Claude Code, Codex.

Per session 74 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,994 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 $0.00074 $0.01994
Opus 5 $0.00037 $0.00997
Sonnet 5 $0.00015 $0.00399
Haiku 4.5 $0.00007 $0.00199

Measured 2d ago against content hash d08b491e7fb9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

implementation-agent 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.

.claude/skills/implementation-agent/SKILL.md · 259 lines

How it starts

The opening of the file, as written. The whole thing — 259 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Implementation Agent

Implements a GitHub issue in a Band workspace. Each run does one piece of work: reads progress from the GitHub issue comment, picks the next task, implements it, commits, pushes, and updates progress. Designed to be called repeatedly — each invocation gets fresh context and picks up where the last one left off.

Prerequisites

  • Inside a Band workspace (git worktree managed by Band)
  • gh CLI authenticated with GitHub
  • band CLI available
  • Branch name encodes the issue number: <issue-number>-<summary> (e.g., 99-task-loop-mode)

Steps

1. Detect Issue and Repo

BRANCH=$(git branch --show-current)
ISSUE_NUMBER=$(echo "$BRANCH" | grep -oE '^[0-9]+')
REPO=$(gh repo view --json nameWithOwner -q .nameWithOwner)

If no issue number is found in the branch name, check the task prompt for a #<number> reference. If still not found, skip issue tracking and just work from the prompt.

Fetch the issue:

gh issue view $ISSUE_NUMBER --repo $REPO

2. Read Progress State

Check for an existing progress comment on the issue:

gh api repos/$REPO/issues/$ISSUE_NUMBER/comments \
  --jq '.[] | select(.body | contains("## Implementation Progress")) | {id, body}'

Also check the workspace state:

git log --oneline main..HEAD       # What's been committed
git status --porcelain              # Any uncommitted work
gh pr list --head $BRANCH --repo $REPO --json number,state,reviewDecision  # Existing PR?

3. Route by Current State

Based on what you find, take the appropriate action:

State A — PR exists and is approved: Go to step 8 (Merge and Clean Up).

State B — PR exists with review feedback: Go to step 7 (Handle PR Feedback).

State C — PR exists, CI failing: Go to step 7 (Handle PR Feedback — fix CI).

State D — PR exists, waiting for review: Nothing to do. Print "PR #N is waiting for review" and exit.

State E — No PR, progress comment exists with unchecked items: Go to step 5 (Implement Next Task).

Read the full file on GitHub · 259 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 · 259 lines · 74 tokens per session scan A d08b491e7fb9

Subscribe to this mod's changes

implementation-agent is a skill published in the GitHub repository band-app/band (5 stars, last pushed 2d ago), licensed MIT. It adds 74 tokens to every session and 1,994 once invoked, about $0.0004 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 2026-08-31.