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/loncadev/baron/agents-mdgit clone --depth 1 https://github.com/loncadev/baronWhat 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 | $0.00529 | $0.00529 |
| Opus 5 | $0.00264 | $0.00264 |
| Sonnet 5 | $0.00106 | $0.00106 |
| Haiku 4.5 | $0.00053 | $0.00053 |
Grade A, and why
baron 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 2d 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.
What it actually says
Work tracking — route through Baron
Track work through Baron, not raw provider writes: it normalizes issues and source control across providers behind one contract, so speak its abstract vocabulary, never a vendor's native states.
- Roles, not native states. Move work by role:
backlog → ready → in_progress → in_review → done, plusblocked. Say "move it to in_progress", never "set the state to Active" — Baron maps the role to the provider. - Type roles this policy maps:
initiative,epic,story,task,bug,subtask. Asking for one it does not map is an error, not a degrade — that list is whatissue.createaccepts here. - Tools:
baron_issue_*(create / get / update / transition / comment / assign / link / query),baron_scm_*(branch / PR),baron_recipe_run, andbaron_learning_*/baron_followup_*for durable decisions and follow-ups. - Daily loop — prefer the skills:
/baron:task-new(create),/baron:task-start <id>(cut the canonical branch, move to in_progress, assign you),/baron:task-finish(draft PR),/baron:task-land(undraft + merge — nevergh/az),/baron:task-move,/baron:task-list,/baron:task-sync. Each item's canonical branch is Baron-derived — use it verbatim, never invent one. - Reading/exploring a provider natively is fine, but make every work-item change through Baron so the role mapping, gap policy, and knowledge loop apply.
On this project (provider: github): roles ride labels (Baron provisions in-progress / in-review / done); sprints are NOT available — sprint queries degrade to empty. That empty is expected here, not a bug;
parent/child is emulated via a parent:<id> label. Where a capability is missing Baron negotiates it (error / emulate /
degrade) and logs it — an empty or emulated result from a degraded capability is expected behavior,
not a silent failure to report as a bug.
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.
- 2d ago First seen · 29 lines · 529 tokens per session scan A bdeab5d7fd7e
baron AGENTS.md is an instructions file published in the GitHub repository loncadev/baron (1 stars, last pushed 12d ago), licensed Apache-2.0. It adds 529 tokens to every session, about $0.0026 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
agent-smith CLAUDE.md
Instructions for holgerleichsenring/agent-smith, covering agent-smith — ai agent instructions, context files (read in this order), phase directory structure, experiential memory (remember / recall) and minting a phase id.
teamxray copilot-instructions.md
Instructions for AndreaGriffiths11/teamxray, covering github copilot instructions for mcp team x-ray extension, response style, build, test, and lint commands, validating extension changes and architecture.
teamxray AGENTS.md
Instructions for AndreaGriffiths11/teamxray, covering agents.md, purpose, tagline, tech stack and architecture.
claude-codex-settings CLAUDE.md
Claude Code instructions for fcakyon/claude-codex-settings, covering claude-settings, repo structure, cross-tool plugin format reference, plugin manifests and root marketplace files.
vscode-gitlens CLAUDE.md
Claude Code instructions for gitkraken/vscode-gitlens, a project described as: Supercharge Git inside VS Code and unlock untapped knowledge within each repository — Visualize code authorship at a glance via Git blame annotations and CodeLens, seamlessly navigate and explore Git repositories, gain valuable insights via…
ghtkn AGENTS.md
Instructions for suzuki-shunsuke/ghtkn, a project described as: A CLI to create short-lived (8 hours) GitHub App User Access Token for secure local development.