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/bartoszarendt/agenticloopWrote 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/bartoszarendt/agenticloop/lifecycle-protocol)<a href="https://agentmods.dev/commands/bartoszarendt/agenticloop/lifecycle-protocol"><img src="https://agentmods.dev/badge/commands/bartoszarendt/agenticloop/lifecycle-protocol/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/bartoszarendt/agenticloop/lifecycle-protocol"><img src="https://agentmods.dev/badge/commands/bartoszarendt/agenticloop/lifecycle-protocol.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 | $0.00000 | $0.01543 |
| Opus 5.5 | $0.00000 | $0.00617 |
| Sonnet 5.5 | $0.00000 | $0.00309 |
| Haiku 4.5 | $0.00000 | $0.00154 |
Grade A, and why
lifecycle-protocol 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 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.
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 — 129 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Lifecycle Protocol
This is the single human-readable lifecycle protocol. AGENTIC_LOOP.md is the
small acting-methodology index; role files describe role boundaries; skills own
procedures; adapters route to those sources. The CLI modules below are the
machine authority. This protocol explains their shared lifecycle without making
prose a mutation input.
Facts, Verdicts, And Refusals
Protected evaluators return a shared fact/verdict shape: ordered facts identify
the canonical fact, observed state, fact owner, policy code when applicable, and
evidence; a verdict is legal, illegal, or unknown; reasons identify failed
or unavailable material facts; prerequisites name required protected inputs.
src/result-envelope.js renders public validation results from that shape.
Every refusal has a stable code from src/refusal-classes.js, a material fact,
a fact owner, and a safe deterministic nextAction when repair is mechanical.
src/repair-policy.js owns routing. A diagnostic does not grant authority. It
cannot move a task, consume a packet, accept review, certify audit, or cross a
human gate. Mechanical recomputation is performed atomically by the protected
CLI command, not delegated to a role.
task explain is the read-only consumer of those evaluator projections. It
returns { derived: true, persisted: false, authority: "none", task, actions };
each action preserves the evaluator's facts, verdict, applicability, reasons,
and prerequisites. It does not create evidence and no receipt or required check
may invoke it as mutation evidence.
Canonical Lifecycle
- Contract and readiness. The Maintainer owns durable task scope and required checks. The CLI validates the trusted task-contract baseline and readiness against current facts. A broad activation scope does not authorize a task and readiness does not create activation.
- Authorization and dispatch.
task prepare-dispatchderives a packet from current task, activation, readiness, repository, backend, role capability, and task-specific evidence. It is a read-only prediction. The packet is stale whenever a protected input changes. - Atomic role start. For files,
task role-start <id> --packet <path>revalidates and consumes the exact packet, records the in-progress carrier transition, and initializes the scratch check aggregate in one transaction. No post-start packet recomputation authorizes work. GitHub uses its guarded backend-equivalent transition. - Implementation evidence. The Engineer modifies only accepted scope,
runs the required checks, and uses CLI-produced execution evidence. On files,
guarded
task evidencemutations append allowed implementation evidence without changing protected task-contract fields. - Return and review.
task prepare-returnderives a raw role return from the consumed packet, current carrier, exact artifact, and check evidence.task verify-returnis the protected receiving boundary. Only a current verified return may enter review preparation; a raw return is not review evidence or completion. - Review, audit, closeout. The Maintainer independently reviews the exact artifact. A fresh Auditor assesses the exact frozen candidate when enabled. Human-controlled closeout consumes the current evidence; no prior green, diagnostic, comment, label, or status auto-crosses the gate.
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 · 129 lines · 0 tokens per session scan A 0aa675412afe
lifecycle-protocol is a command published in the GitHub repository bartoszarendt/agenticloop (2 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,543 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 2026-09-13.
Other commands, from other repositories
clean
Clean up old Answer me with HTML pages, videos and the narration cache.
config
View or change Answer me with HTML settings — auto-open browser, default theme, light/dark, STE strictness, video voice, update notices.
dashboard-session-fork
Fork a session into a new one. Usage /dashboard:session-fork.
dashboard-session-hide
Hide a session from the dashboard list. Usage /dashboard:session-hide.
flow-nexus-sandbox
E2B sandbox management for isolated code execution.
setup
Command "setup" from ruvnet/flow-nexus, covering setting up ruv-swarm hooks, quick start, 1. initialize with hooks, 2. test hook functionality and test pre-edit hook.