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/codingriver/upilot/agents-mdgit clone --depth 1 https://github.com/codingriver/upilotWhat 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 | $0.00729 | $0.00729 |
| Opus 5 | $0.00365 | $0.00365 |
| Sonnet 5 | $0.00146 | $0.00146 |
| Haiku 4.5 | $0.00073 | $0.00073 |
Grade A, and why
upilot 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.
How it starts
The opening of the file, as written. The whole thing — 28 lines — stays where its author put it; the contents beside it link to each section on GitHub.
UPilot Repository Instructions
Release Workflow
- Treat commands such as
发布版本 v0.3.22,并检查打包结果和版本号是否正确,发布 v0.3.22,触发发布版本 0.3.22, and semantically equivalent requests as explicit authorization to run the GitHub Actions workflow named发布版本from.github/workflows/prepare-release.yml. - Use the requested semantic version after normalizing it to
vMAJOR.MINOR.PATCH. If the user requests the next version without specifying a number, inspect existing tags and use the next patch version unless the user states otherwise. - Trigger the workflow from
mainwithgh workflow run prepare-release.yml --ref main -f version=<version>. Do not manually create or push the release tag, and do not manually maintain release versions inpackage.jsonorupilotserver~/pyproject.toml; the workflow owns those operations. - Before triggering a release, verify that the intended release changes are committed and pushed to
main. Do not include unrelated local changes without explicit authorization. - After triggering, monitor both the
发布版本run and its downstreamBuild UPilot MCP Server EXErun until completion. A successful dispatch alone is not a successful release. - Verify that the release tag exists, the GitHub release and expected assets were published, and the version in the tag's
package.jsonandupilotserver~/pyproject.tomlexactly matches the requested version. Report run URLs, commit/tag information, asset results, and any version mismatch or failed job clearly. - If either workflow fails or any version is inconsistent, do not report release success. Preserve the failure logs and identify the failed step and corrective action.
Testing And Validation
- Use
./Tests~/UPilotTestas the default and canonical Unity project for UPilot package compile and EditMode acceptance. - After UPilot C# or assembly-related changes, validate against
./Tests~/UPilotTestbefore claiming Unity compile/EditMode acceptance, unless the user explicitly says not to run tests. - Do not attempt a full EditMode suite by default. Unless the user explicitly requests a full/complete EditMode regression, run only the narrowest targeted tests that cover the changed code and its direct regressions.
- A full EditMode run is authorized only when the user explicitly asks for full, complete, regression, acceptance, or equivalent whole-suite validation; otherwise report targeted results and any unrun coverage.
- If a full-suite run is not explicitly authorized, do not start it speculatively after targeted tests pass, even when preparing a final handoff.
- Do not use external client projects such as
D:\MA\xclientorF:\xclient2as default UPilot validation projects. - Use external client projects only when the user explicitly requests project-side/business smoke validation or investigation.
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.
- yesterday First seen · 28 lines · 729 tokens per session scan A 332cfd128f30
upilot AGENTS.md is an instructions file published in the GitHub repository codingriver/upilot (0 stars, last pushed 4d ago), licensed MIT. It adds 729 tokens to every session, about $0.0036 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
unity-mcp CLAUDE.md
Claude Code instructions for CoplayDev/unity-mcp, covering claude.md, what this project is, architecture, three layers on the python side and transport modes.
mcp-unity AGENTS.md
Instructions for CoderGamester/mcp-unity, covering mcp unity — ai agent guide (mcp package), purpose (what this repo is), how it works (high-level data flow), key defaults & invariants and repo layout (where to change what).
roblox-brain AGENTS.md
AGENTS.md instructions for TabooHarmony/roblox-brain, covering agents.md: guide for ai agents and contributors, what this repo is, architecture: progressive disclosure, skill file structure and skill title.
rails-ai-context AGENTS.md
Instructions for crisnahine/rails-ai-context, covering rails-ai-context, agent skills, issue tracker, triage labels and domain docs.
cyrus CLAUDE.md
Claude Code instructions for cyrusagents/cyrus, covering claude.md, project overview, how cyrus works, example interaction and test drives.
bloxforge CLAUDE.md
Instructions for princeofscale/bloxforge, covering bloxforge agent guide, project purpose, repository map, invariants and working rules.