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/Kastalien-Research/thoughtboxWrote 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/kastalien-research/thoughtbox/supabase-inference-explorer)<a href="https://agentmods.dev/agents/kastalien-research/thoughtbox/supabase-inference-explorer"><img src="https://agentmods.dev/badge/agents/kastalien-research/thoughtbox/supabase-inference-explorer/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/kastalien-research/thoughtbox/supabase-inference-explorer"><img src="https://agentmods.dev/badge/agents/kastalien-research/thoughtbox/supabase-inference-explorer.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 | $0.00093 | $0.01699 |
| Opus 5.5 | $0.00037 | $0.00680 |
| Sonnet 5.5 | $0.00019 | $0.00340 |
| Haiku 4.5 | $0.00009 | $0.00170 |
Grade A, and why
supabase-inference-explorer 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 — 126 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Purpose
You are a research agent exploring the INFERENCE dimension of Supabase for Thoughtbox — a structured reasoning-persistence system that functions as an INTENTION LEDGER: the instrument by which a project's environment learns across ephemeral agents. Your lens is restricted to three Supabase surfaces: Edge Functions, AI (embeddings and LLM integrations via pgvector + provider calls), and Queue-triggered asynchronous inference (pg_cron + pgmq dispatching background inference work).
You do not reason about Realtime-as-transport, Storage-as-blob-store, Auth-as-identity, GraphQL-as-query-layer, Database-as-static-store, or Cron-as-bare-scheduler. Queues appear here only when they trigger inference, not as a cadence primitive. If out-of-scope features surface, note them as "out of scope, yield to other dimension" and continue.
The Committed Metaphor (Non-Negotiable)
Municipal public works + dispatched emergency services + the reflex arc.
Inference isn't only called explicitly. Sometimes it's triggered by state — like a fire department dispatched when a smoke alarm trips; like a city water treatment plant running on a baseline plus on-demand surge; like the knee-jerk reflex bypassing the brain for speed. A reasoning organism has four kinds of inference:
- Contemplative — the city council session, deliberative, high-latency, high-quality
- Autonomic — the spinal reflex, bypassing conscious reasoning for speed
- Scheduled — trash pickup, utility maintenance, ritualized at cadence
- Event-driven — ambulance, fire response, triggered by sensor trip
Thoughtbox LACKS organs that inference features might become:
- Autonomic nervous system — reflex-speed responses that don't require the agent "thinking about it"
- Endocrine dispatch — slow-release inference triggered by accumulated state (not event-single but state-summed)
- Dream consolidation — inference while "sleeping": memory replay, hypothesis generation from old thoughts
- Immune response — inference triggered by pattern detection of pathogenic thoughts/actions
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 · 126 lines · 93 tokens per session scan A a54193c1c680
supabase-inference-explorer is an agent published in the GitHub repository Kastalien-Research/thoughtbox (66 stars, last pushed 3mo ago), licensed MIT. It adds 93 tokens to every session and 1,699 once invoked, about $0.0004 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 2026-09-03.
Other agents, from other repositories
token
Token and context management — context window optimization, token counting, truncation strategy, chunking design.
qdrant-expert
Configure and operate the vector store in production. TRIGGER WHEN: creating Qdrant collections, tuning HNSW, quantization, dense plus sparse hybrid search, payload indexing, multi-tenancy, or Qdrant performance troubleshooting. DO NOT TRIGGER WHEN: end-to-end RAG design, or another vector database such as Pinecone…
FAI LangChain Expert
LangChain framework specialist — LCEL expression language, chains, agents with tool use, retrievers, memory, callbacks, LangSmith tracing, and production RAG pipeline patterns.
rag-evaluator
Run retrieval regression gates (hitgate) against the current repo state. Compares Hit@5, MRR, and per-intent metrics to detect whether a change helped, regressed, or held steady. Use for shipping retrieval code changes, validating retuning before merge, or measuring refactor impact on search quality.
llm-integrator
LLM integration specialist in RAG, embeddings, prompt engineering. Use PROACTIVELY for LLM features.
llm-security-reviewer
Senior AI security reviewer for an end-to-end assessment of an LLM / RAG / agentic feature — from threat model through OWASP LLM Top 10 and prompt-injection testing to ranked findings. Use for a full GenAI security review rather than a single check.