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.
git clone --depth 1 https://github.com/gf-labs/claude-toolboxWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/gf-labs/claude-toolbox/recap)<a href="https://agentmods.dev/commands/gf-labs/claude-toolbox/recap"><img src="https://agentmods.dev/badge/commands/gf-labs/claude-toolbox/recap/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/gf-labs/claude-toolbox/recap"><img src="https://agentmods.dev/badge/commands/gf-labs/claude-toolbox/recap.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00063 | $0.01687 |
| Opus 5.5 | $0.00025 | $0.00675 |
| Sonnet 5.5 | $0.00013 | $0.00337 |
| Haiku 4.5 | $0.00006 | $0.00169 |
Grade A, and why
recap scanned grade A with 1 finding 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 1mo 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.
Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
result = subprocess.run(['git', 'log', '--oneline', '--since', since], capture_output=True, text=True) How it starts
The opening of the file, as written. The whole thing — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Arguments
$ARGUMENTS
Parse --days N or --hours N from arguments. Default: current session (no time window). If --hours N is given, convert to fractional days (hours / 24). --days takes precedence over --hours if both are given.
Determine mode before running any commands:
- If no
--daysor--hoursargs → session mode - If
--daysor--hourspresent → time-window mode
Collect context
Run all commands now before producing output.
Scope + date:
python3 ${CLAUDE_TOOLBOX_ROOT}/scripts/_scope.py
date +%Y-%m-%d
Current session (always run):
python3 ${CLAUDE_TOOLBOX_ROOT}/scripts/collect-summarize.py
Git branch + uncommitted state:
git branch -vv 2>/dev/null | head -1
git status --short 2>/dev/null | head -10
Git commits — time-window mode only (skip if session mode):
python3 -c "
from datetime import datetime, timedelta
import subprocess, sys
args = '$ARGUMENTS'.split()
days = None
for i, a in enumerate(args):
if a == '--days' and i+1 < len(args):
days = int(args[i+1]); break
elif a == '--hours' and i+1 < len(args):
days = int(args[i+1]) / 24; break
if days is None:
print('(session mode — see git state above)')
sys.exit(0)
since = (datetime.now() - timedelta(days=days)).strftime('%Y-%m-%dT%H:%M:%S')
result = subprocess.run(['git', 'log', '--oneline', '--since', since], capture_output=True, text=True)
print(result.stdout or '(no commits in window)')
"
Files changed — time-window mode only (skip if session mode; session mode uses FILES_TOUCHED above):
python3 -c "
from datetime import datetime, timedelta
import subprocess, sys
args = '$ARGUMENTS'.split()
days = None
for i, a in enumerate(args):
if a == '--days' and i+1 < len(args):
days = int(args[i+1]); break
elif a == '--hours' and i+1 < len(args):
days = int(args[i+1]) / 24; break
if days is None:
print('(session mode — see FILES_TOUCHED above)')
sys.exit(0)
since = (datetime.now() - timedelta(days=days)).strftime('%Y-%m-%dT%H:%M:%S')
result = subprocess.run(['git', 'log', '--name-only', '--pretty=format:', '--since', since], capture_output=True, text=True)
files = sorted(set(f for f in result.stdout.splitlines() if f.strip()))
print('\n'.join(files) if files else '(none)')
"
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.
- 1mo ago First seen · 164 lines · 63 tokens per session scan A db6c658d8515
recap is a command published in the GitHub repository gf-labs/claude-toolbox (2 stars, last pushed 1mo ago), licensed MIT. It adds 63 tokens to every session and 1,687 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
ccc-parallel
Run several Claude Code agents at once, each in its own git worktree and branch, then merge the results through one coordinator. Use when you have independent tasks to develop concurrently.
ccc-pr
Open a pull request for the current branch: commit pending work, run tests, push, and create the PR with gh. Use when a branch is ready for review.
catchup
Restore context after /clear by summarizing recent work and project state.
commit
Split the changes into atomic commits, preflight and push them. Pass --skip-preflight or --no-push to skip a step.
quick
Quick execution mode for ad-hoc tasks without full workflow overhead.
ship
Branch, commit, open PR, gather Claude + every enabled AI reviewer (Copilot, CodeRabbit, etc.), fix/justify/resolve every finding, loop until clean, then merge. Run only when implementation is finished AND the owner has said to ship (e.g. "ship it") — never self-invoke just because the work looks done. To design and…