Borrowing it
Nothing to install: this file belongs to sortie-ai/sortie. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/sortie-ai/sortie/main/AGENTS.mdgit clone --depth 1 https://github.com/sortie-ai/sortieWrote 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/sortie-ai/sortie/agents-md)<a href="https://agentmods.dev/instructions/sortie-ai/sortie/agents-md"><img src="https://agentmods.dev/badge/instructions/sortie-ai/sortie/agents-md/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/instructions/sortie-ai/sortie/agents-md"><img src="https://agentmods.dev/badge/instructions/sortie-ai/sortie/agents-md.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 | $0.01420 | $0.01420 |
| Opus 5.5 | $0.00568 | $0.00568 |
| Sonnet 5.5 | $0.00284 | $0.00284 |
| Haiku 4.5 | $0.00142 | $0.00142 |
Grade A, and why
sortie 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 3d 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 — 56 lines — stays where its author put it; the contents beside it link to each section on GitHub.
sortie
An agent orchestrator that ships as one statically-linked binary with no runtime dependencies. Most of its structure is enforced by convention rather than by the compiler: go build accepts every layering violation listed below.
Commands
make help lists the build, test, and lint targets; they carry the flags CI uses, so prefer one when it fits. Calling the Go toolchain directly is fine when no target does what you need.
Gotchas
- Go resolves through the asdf shim at
~/.asdf/shims/go. OverridingGOPATH,GOMODCACHEor the binary path breaks the pinned toolchain;GOFLAGSis not such an override and is allowed. - Symphony is prior art, not a template. Sortie derives from OpenAI Symphony but diverges intentionally. Do not port Symphony patterns or Elixir idioms.
- Generic naming in core code. Use
agent_*,tracker_*,session_*in orchestrator core. Neverjira_*,claude_*,codex_*outside their adapter packages. - The orchestrator resolves adapters by
kindthroughinternal/registry. Importing a package that registers a kind in itsinit()is a layering violation there even thoughgo buildaccepts it; a package alongside it that registers none may be imported directly.cmd/sortieblank-imports every kind package solely to trigger that registration. internal/scm/is an adapter family. Apply the same boundary rules asinternal/tracker/*/: no cross-adapter imports, must not import the orchestrator, normalize external responses to domain types at the boundary.- Shared adapter helpers go in an
internal/package named for the concern it serves, never duplicated per adapter and never ininternal/domain/. Logic scoped to one adapter family nests under that family instead. - Integration tests are env-gated by a per-adapter
SORTIE_<ADAPTER>_TEST=1. Without the gate set they must skip cleanly — never fail. - Tests wait on events, never on a clock. A test stops the code after its readiness or completion signal (for
run(),runUntilStartedon thestartupDonehook), never by handing it a short deadline, and proves a fast path by its result or undertesting/synctest, not by timing it. Every wait goes throughinternal/waittest:HangGuardis the one hang guard, and a live suite passes its measured bound to the same waits. Notime.Sleepfor ordering. A raised timeout or a rerun job does not fix a flake. workflow.Manager.Reload()is fail-safe. On parse or validation error the previouscurrentConfigandcurrentPromptare retained andLastLoadError()reports the failure. Preserve that invariant; neveros.Exiton a bad WORKFLOW.md.- Prompt templates render with
Option("missingkey=error"). Adding a template variable without wiring its data field is a runtime error, not an empty string. - The dashboard draws structure with layout, never with characters. No Unicode arrows, check marks, bullets or box glyphs to connect or mark values (
a → b,✓ done), and no·separator where a column, a second line, a chip or a stepper can carry the split. Glyphs come fromweb/src/components/icons.tsx. - The dashboard's build output under
internal/server/ui/distis committed.make ui-buildregenerates it fromweb/; commit it with the sources it was built from, and resolve a conflict in it by rebuilding, never by hand.
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.
- 3d ago Changed · +1 lines · +17 tokens per session 97939c50c9bb
- 4d ago Changed · +1 lines · +129 tokens per session 84fd12b26d50
- 7d ago Changed · +40 tokens per session 4ae299962d52
- 14d ago Changed · +1 lines · +78 tokens per session 5a7f52dfab22
- 15d ago Changed · +1 lines · +54 tokens per session dc4aac14b5d8
- 29d ago Changed · +1 lines · +128 tokens per session 82ce17465d21
- 1mo ago First seen · 51 lines · 974 tokens per session scan A c70741dcf2c1
sortie AGENTS.md is an instructions file published in the GitHub repository sortie-ai/sortie (209 stars, last pushed today), licensed Apache-2.0. It adds 1,420 tokens to every session, about $0.0057 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 2026-09-20.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode copilot-instructions.md
Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).
deepseek-harness AGENTS.md
AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-stable apis and released session data, repository layout, commands and host sandbox failures.