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 instructions/liuyingxuvka/flowguard/agents-mdgit clone --depth 1 https://github.com/liuyingxuvka/FlowGuardWrote 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/instructions/liuyingxuvka/flowguard/agents-md)<a href="https://agentmods.dev/instructions/liuyingxuvka/flowguard/agents-md"><img src="https://agentmods.dev/badge/instructions/liuyingxuvka/flowguard/agents-md.svg" alt="Measured on agentmods" 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 | $0.02987 | $0.02987 |
| Opus 5 | $0.01494 | $0.01494 |
| Sonnet 5 | $0.00597 | $0.00597 |
| Haiku 4.5 | $0.00299 | $0.00299 |
Grade A, and why
FlowGuard 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 5d 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 — 156 lines — stays where its author put it; the contents beside it link to each section on GitHub.
FlowGuard Project Rules
For non-trivial work, select the smallest current FlowGuard public owner; clear satellites are direct peers and unclear ordinary behavior/state work uses flowguard.
FlowGuard repository: https://github.com/liuyingxuvka/FlowGuard
FlowGuard agent skill suite: Primary agent surface: the current clean consumer projection at $CODEX_HOME/skills/flowguard/SKILL.md; the project does not copy the FlowGuard suite into its local tree, and the Python package/CLI is not the AI-agent skill installation surface.
Project record: .flowguard/project.toml; machine log: .flowguard/adoption_log.jsonl; human log: docs/flowguard_adoption_log.md.
Current adoption record: FlowGuard check-engine version: 0.68.16; FlowGuard schema version: 1.0.
Before non-trivial work run python -m flowguard project-audit --root .; if the installed engine is newer, run full project-upgrade scanning and affected revalidation, and if older connect the current engine.
Use latest-schema-first direct replacement; obsolete fields, aliases, wrappers, and alternate success paths have no normal-runtime fallback.
Only the sole content-addressed observed_implementation head is current; targets, experiments, discovery, and green candidates do not own current behavior.
Change model authority only through one accepted ModelRevisionSet; keep the revision-local delta distinct from its complete CurrentEffectiveIntentView, bind every current model owner exactly, persist evidence before the pointer, and restore/compensate effects before rollback.
Default replacement means dispose the old path: every replaced field, alias, wrapper, or alternate success needs an explicit delete/block/migrate/delegate/repair/replace/scope disposition.
Broad behavior claims require an independent BehaviorCommitmentLedger inventory, one plane and primary owner per admitted promise, and Primary Path Authority for path-sensitive rows.
Commitments stay in product_runtime, agent_operation, or development_process; select a bounded same-plane owner closure and keep related planes as typed context only.
Declare ledger mode first; only bootstrap/backfill may discover broadly, while add/change/remove/miss stays on the affected commitment closure.
Field-bearing work uses FieldLifecycleMesh and accounts owner, readers/writers, projection, lifecycle evidence, and old-field disposition.
UI runnable claims and file/work-package claims need current real-surface or payload evidence before broad confidence.
Commitments with path_sensitive=true need one Primary Path Authority, visible primary failure, no alternate automatic success, and current exhaustion/test/risk evidence.
One exact purpose has one intent, commitment, and primary path; equivalent UI/API/CLI/adapter/wrapper surfaces delegate rather than own independent success.
UI Flow Structure owns product language and complete rendered control/display/transition/overlay/recovery/blindspot coverage for full UI claims.
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.
- 5d ago First seen · 156 lines · 2,987 tokens per session scan A 65cbe893dec1
FlowGuard AGENTS.md is an instructions file published in the GitHub repository liuyingxuvka/FlowGuard (2 stars, last pushed 10d ago), licensed MIT. It adds 2,987 tokens to every session, about $0.0149 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 instructions, from other repositories
openscreen AGENTS.md
AGENTS.md instructions for getopenscreen/openscreen, covering agents.md, setup commands, development principles, project layout and code style.
commonly CLAUDE.md
Claude Code instructions for Team-Commonly/commonly, covering claude.md / agents.md, 🧠 product vision & architecture philosophy, what commonly is, the architecture model and key concepts.
Claw3D AGENTS.md
Instructions for iamlukethedev/Claw3D, covering agent instructions, cursor cloud specific instructions, service overview, running the app and lint, typecheck, and tests.
claude-ads CLAUDE.md
Instructions for AgriciDaniel/claude-ads, covering claude ads repository instructions, architecture, development and verification.
openstatus AGENTS.md
AGENTS.md instructions for openstatusHQ/openstatus, covering agents.md, verify your change, toolchain, architecture and tests.
openstatus CLAUDE.md
Claude Code instructions for openstatusHQ/openstatus, covering claude.md and investigating production with polylane.