temple8 AGENTS.md

Repository instructions for temple8, a Python project whose work moves through defined stages using flowr, its single routing tool. They describe how agents receive tasks, read input files, create output files, and provide verified evidence.

In plain words
What is it for?
Use them when implementing or coordinating staged tasks, dispatching agents, handling workflow artifacts, or asserting transition evidence.
Why use it?
They prevent skipped stages, improvised routing, missing inputs, and unsupported claims about completed work.

Instructions file for CodexOpenCode

Install

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.

agentmods
npx agentmods add instructions/nullhack/temple8/agents-md
Clone the repo
git clone --depth 1 https://github.com/nullhack/temple8

Made for: Codex, OpenCode.

Per session 2,589 This file is loaded in full into every session.
When invoked 2,589 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.02589 $0.02589
Opus 5 $0.01295 $0.01295
Sonnet 5 $0.00518 $0.00518
Haiku 4.5 $0.00259 $0.00259

Measured yesterday against content hash 5cd812fdd3b3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

temple8 AGENTS.md 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.

AGENTS.md · 153 lines

How it starts

The opening of the file, as written. The whole thing — 153 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md

A Python project with the staged-contract workflow wired in (flow set, agents/skills/knowledge, CI, tooling). Drive the pipeline one state at a time through flowr.

Binding constraints

  1. flowr is the single router. Every state change runs through flowr — no improvised routing, no skipping, no pre-empting a later state's work. One state's work is exactly its output artifacts; the next begins only when the orchestrator transitions.
  2. The orchestrator dispatches; the dispatched agent does the work. The orchestrator MUST invoke the state's dispatch_to agent — it never authors the work. The handoff gives the agent: skills paths, input artifacts, output artifacts (write only here), and the evidence keys to assert. The agent produces the artifacts and asserts evidence; it does not transition.
  3. The state's contract is binding. Read every input artifact before starting — missing means stop, not assume. Write only to output artifacts.
  4. Assert only verified evidence; CI is the backstop. On a guarded transition flowr fires only on the dispatched agent's asserted evidence (--evidence k=v); assert nothing you did not check — CI catches the lie (ruff / pyright / mypy.stubtest / pytest). Unguarded transitions (discovery, explore) carry no flowr gate — the orchestrator verifies the output artifacts and the stakeholder's approval IS the gate.
  5. Branch discipline. Match the state's git branch: discovery/explore/plan/deliver run on dev; build runs on feature/<session_id>, cut from dev at build entry. The contract surface is committed to dev at plan; the feature branch carries only the source implementation. Squash-merge feature/<session_id>dev only under the whole-suite gates, then delete the branch. No dangling branches.
  6. Every requirement traced. Each interview finding maps to a test or an explicit deferral; an untraced requirement is a gap the simulate gate rejects.
  7. Only the orchestrator transitions. Producing artifacts is not finishing — the orchestrator's verified flowr transition is. A dispatched agent never runs flowr transition / flowr session, never assumes the next state, never declares the flow done.
  8. Todo-first. The first action on entering a state is to create its one-state todo (the loop below). No todo = no work. Only the orchestrator holds the todo; regenerate it after every transition.

Read the full file on GitHub · 153 lines

Changes

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.

  1. yesterday First seen · 153 lines · 2,589 tokens per session scan A 5cd812fdd3b3

Subscribe to this mod's changes

temple8 AGENTS.md is an instructions file published in the GitHub repository nullhack/temple8 (11 stars, last pushed 27d ago), licensed MIT. It adds 2,589 tokens to every session, about $0.0129 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 2026-08-30.