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/MadaraUchiha-314/the-loopWrote 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/madarauchiha-314/the-loop/finish-tasks)<a href="https://agentmods.dev/commands/madarauchiha-314/the-loop/finish-tasks"><img src="https://agentmods.dev/badge/commands/madarauchiha-314/the-loop/finish-tasks/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/madarauchiha-314/the-loop/finish-tasks"><img src="https://agentmods.dev/badge/commands/madarauchiha-314/the-loop/finish-tasks.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 | $0.00023 | $0.00839 |
| Opus 5.5 | $0.00009 | $0.00336 |
| Sonnet 5.5 | $0.00005 | $0.00168 |
| Haiku 4.5 | $0.00002 | $0.00084 |
Grade A, and why
finish-tasks 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 18d 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.
How it starts
The opening of the file, as written. The whole thing — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.
the-loop: finish-tasks $ARGUMENTS
Wrap up a work item once all tasks are complete and reviewed. This is the cleanup step of the loop; the cleanup set is intentionally extensible (more may be added later).
Read the the-loop skill and reference/workflow.md. Load .the-loop/harness-config.yaml.
Steps
-
Verify done. Resolve
$ARGUMENTStodocs/specs/<id>/. Confirm every task intasks.mdis checked (- [x]), every PR inevidence/pull-requests.md's table is merged or closed (a work item may be delivered by several), the required self/critic reviews and human review are recorded, and validated evidence has been presented. If anything is outstanding, stop and report what remains (point back to/the-loop:execute-tasks <id>). -
Mark complete. Set the ticket phase label to
loop:complete; the outcome and its evidence are already recorded where each gate left them, underevidence/. -
Post the completion summary — first, and promptly. One comment on the ticket, posted with
the-loop comment --work-item <ref> --body-file -(it stamps the marker and reaches the room through the mirror; issue-447): what shipped, the evidence links, the accepted gaps. This is the endgame's one message a reader most wants, and it is raced by the close (issue-405 P2): when the merge'sCloses #Nhas already closed the ticket, the daemon holds the session forrouting.tmux.finishGraceSeconds(default 300 s) waiting for exactly this summary and the claim below, then ends it. Post the summary before anything else here; do no other work in between. -
Claim completion.
the-loop graph complete <id>— the claim that tells the graph, and through it the daemon, that the terminal node's work is done. The daemon reads it as "the session has finished" and ends the held closure at once, so the pane, the checkout and the record go the moment you are done rather than at the deadline. -
Cleanup (extensible):
- Close the ticket(s) for this work item (GitHub issue / Jira) if still open,
with
the-loop ticket close <ref>(issue-447;--reason not_plannedfor work that was dropped), after athe-loop commentthat references the merged PR(s) and evidence. On ajira:ref it transitions the ticket into Done; when the ticket offers no such transition, or several, nothing moves and the error lists them — escalate withthe-loop asknaming them, or have the operator setintegrations.jira.closeTransition. The Jira MCP tools only when the CLI is not installed. A ticket a merge'sCloses #Nalready closed is expected — say so in the summary and move on; do not reopen it. Closing the ticket is also what ends the work item's harness session — one of its PRs merging does not (seereference/automation.md). - Future cleanup steps are added here (e.g. archiving branches, releasing artifacts, notifying channels). Keep this list the single place cleanup grows.
- Close the ticket(s) for this work item (GitHub issue / Jira) if still open,
with
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.
- 18d ago Changed · +4 lines 6ad929cac9e5
- 24d ago Changed · +3 lines f0a91276cf3e
- 1mo ago Changed · +16 lines a2ba9de68a38
- 1mo ago Changed · -1 lines 1a40ba69b5c2
- 1mo ago Changed 2225581dca8e
- 1mo ago First seen · 35 lines · 23 tokens per session scan A 8b9ff5c13a80
finish-tasks is a command published in the GitHub repository MadaraUchiha-314/the-loop (7 stars, last pushed 2d ago), licensed MIT. It adds 23 tokens to every session and 839 once invoked, about $0.0001 per session on Opus 5.5. 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.
Other commands, from other repositories
close
They operate it without you.
okr
Create or review OKRs (Objectives and Key Results).
project-health
Assess overall project health across code quality, tests, docs, and activity.
retro
Run a sprint retrospective with data-driven insights from git history.
sprint-health
Check current sprint health and identify risks.
sprint-plan
Plan a sprint by analyzing backlog, estimating capacity, and selecting work.