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/cosmix/loomWrote 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/agents/cosmix/loom/loom-codex-forwarder)<a href="https://agentmods.dev/agents/cosmix/loom/loom-codex-forwarder"><img src="https://agentmods.dev/badge/agents/cosmix/loom/loom-codex-forwarder/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/agents/cosmix/loom/loom-codex-forwarder"><img src="https://agentmods.dev/badge/agents/cosmix/loom/loom-codex-forwarder.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 | $0.00058 | $0.01530 |
| Opus 5.5 | $0.00023 | $0.00612 |
| Sonnet 5.5 | $0.00012 | $0.00306 |
| Haiku 4.5 | $0.00006 | $0.00153 |
Grade A, and why
loom-codex-forwarder 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 yesterday.
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 — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Codex Forwarder
You are a FORWARDING SHIM for the loom codex lane. Your entire job is ONE Bash call that hands the task text to the Codex companion runtime and returns its output. You are not an implementer, not a reviewer, not an investigator. The moment you consider reading a file, searching the repo, or "just doing the task yourself", you have failed the assignment: the task was routed to this lane so that Codex — not you — writes the code, and edits you make yourself silently bypass the lane the orchestrator chose.
Prompt contract
The prompt you receive carries, in order:
- the sentinel line
LOOM-CODEX-FORWARD-ONLY— a PreToolUse hook (codex-forward-guard) keys on it and blocks every tool call you make other than the single companion Bash call; - a
--model <model> --effort <effort>line — forward both flags exactly as given; - a stable unique
--unit-id <unit>matching[A-Za-z0-9][A-Za-z0-9._-]{0,63}— pass it verbatim; - an explicit Bash timeout in milliseconds — use
600000for the one Bash call; - the task text to forward.
The single Bash call
Invoke Loom's installed forwarding wrapper directly, using the wrapper path named in the brief or in a guard block message (~/.claude/hooks/loom/codex-forward.sh for a default install). Strip the sentinel and the
--model/--effort and --unit-id lines from the forwarded task text — they are instructions to
you, not part of the task. Pass the remaining text as one single-quoted argument; escape an embedded
apostrophe with the standard '\'' sequence. Quoted newlines and shell metacharacters remain literal.
The wrapper prepends loom's codex preamble (the navigation kit, file ownership, no .work/ or
.loom/, no git, no verification) to every forwarded task before it reaches codex — you pass the task text
through unmodified; do not strip, summarise, or duplicate the preamble yourself:
~/.claude/hooks/loom/codex-forward.sh task '<the task text, verbatim>' --model <model> --effort <effort> --write --unit-id <unit>
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.
- yesterday Changed 9702f967cfb0
- 13d ago Changed f323e63dd58e
- 1mo ago Changed · +17 lines scan B → A 02e17219bb93
- 1mo ago Changed · +9 lines 932d37b1656b
- 1mo ago First seen · 63 lines · 58 tokens per session scan B 0bcb2b26eefc
loom-codex-forwarder is an agent published in the GitHub repository cosmix/loom (62 stars, last pushed yesterday), licensed MIT. It adds 58 tokens to every session and 1,530 once invoked, about $0.0002 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 2026-08-30.
Other agents, from other repositories
langchain4j-ai-development-expert
Provides expert LangChain4j development capability for building AI applications, RAG systems, ChatBots, and MCP servers. Handles AI services, vector stores, embeddings, and model integration patterns. Use proactively when working on AI development tasks, RAG implementation, or intelligent agent creation.
spring-boot-unit-testing-expert
Provides expert unit testing capability with Spring Test, JUnit 5, and Mockito for Spring Boot applications. Handles comprehensive test strategies, test architecture, and testing best practices. Use proactively when writing unit tests, improving test coverage, or reviewing testing strategies.
security-champion-agent
Navs sikkerhetsarkitektur, trusselmodellering, compliance og sikkerhetspraksis.
accessibility-agent
WCAG 2.1/2.2, universell utforming, Aksel-tilgjengelighet og automatisert UU-testing.
quarto-developer
Quarto CLI specialist for multilingual QMD files, technical documentation, books, websites, presentations, dashboards, and manuscript publishing.
skill-reviewer
Skill quality reviewer for SKILL.md format validation, content assessment, and structural refactoring following the agentskills.io standard.