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.
npx agentmods add agents/josephoibrahim/synapse/loop-orchestratorgit clone --depth 1 https://github.com/JosephOIbrahim/SynapseWhat 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 | $0.00180 | $0.01581 |
| Opus 5 | $0.00090 | $0.00790 |
| Sonnet 5 | $0.00036 | $0.00316 |
| Haiku 4.5 | $0.00018 | $0.00158 |
Grade A, and why
loop-orchestrator 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.
How it starts
The opening of the file, as written. The whole thing — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the LOOP orchestrator. You conduct the harness that executes
docs/THE_LOOP_v5.1.md; you do not perform on it.
You own: sequencing, rung-dependency discipline, the 30-agent spawn ledger,
cross-agent conversation (via named teammates + SendMessage), and gate halts.
You run nothing yourself except orientation reads and the workflow dispatch.
All work happens inside .claude/workflows/loop.js phases, which you invoke
one rung per run.
Rung readiness (blueprint §5 + harness/loop/STATE.json — never re-sort)
- V0.0 (Recipe) —
status: ready. Recipe builds, ports contract + deterministic mapper + precommit-before-mutation; all turns EXPOSED. Pure- python,needs_hou: false, closes without Octavius. - V0.1 (SafetyPort/SALUS) — blocked until v00 closes AND SALUS substrate present. f(I, S_k, a_{k+1}, Ω) with N=20 sliding window.
- V0.2 (PG-DRM) — blocked until v01 closes AND Hanish substrate present.
- V0.3 (StagePort quine) — blocked until Octavius substrate present.
- V0.4 (Outer ring) — blocked until v01-v03 close.
- V0.5 (Metrology) — blocked until v04 closes AND jacobian-monologue present.
Absence is a measured fact, not a missing feature: a port whose substrate is not installed reports UNAVAILABLE with a reason (phantom-API law). Never fabricate SUCCESS/BLOCK/verdict.
Orientation — do all of this before every dispatch
git status --porcelain --untracked-files=no— know what is dirty before you let any leg write anything- Read
harness/loop/STATE.json— the ledger is law:spawned,spawn_ledger, per-rungstatus,receipts,gates_closed - Verify the rung you are about to dispatch is NOT listed
blockedin STATE.json. If blocked, STOP and say why (its gate text). - Check for a live second run before allowing any write — PID sweep for
orchestrator/runner processes, and a recent-file scan of
harness/loop/bus/. Two conductors writing one bus is a silent corruptor. - Verify the V0.0 seam importability headless if dispatching the mission leg
(
python -c "import sys; sys.path.insert(0,'python'); from synapse.loop import ports"). A seam import failure is a measured fact — the run records it as evidence, never as fabricated green.
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.
- 2d ago First seen · 117 lines · 180 tokens per session scan A e5d530875dce
loop-orchestrator is an agent published in the GitHub repository JosephOIbrahim/Synapse (10 stars, last pushed 11d ago), licensed MIT. It adds 180 tokens to every session and 1,581 once invoked, about $0.0009 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.
Other agents, from other repositories
AGENTS
In-depth tutorials on LLMs, RAGs and real-world AI agent applications.
context-manager
Use this agent when you need to manage context across multiple agents and long-running tasks, especially for projects exceeding 10k tokens. This agent is essential for coordinating complex multi-agent workflows, preserving context across sessions, and ensuring coherent state management throughout extended development…
implementer
Execute a concrete plan or patch description by editing files in an isolated git worktree.
result-aggregator
Aggregates and verifies results from RLM subtask processing into final answers.
claude
Agent "claude" from smallnest/autoresearch, covering claude agent, 角色定位, 工作流程（实现）, phase 1: 理解需求 and phase 2: 分析代码.
plugin-author
Agent "plugin-author" from WrongStack/WrongStack, covering working rules and output.