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/canyonroad/agentsh/agents-mdgit clone --depth 1 https://github.com/canyonroad/agentshWhat 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 | $0.00433 | $0.00433 |
| Opus 5 | $0.00217 | $0.00217 |
| Sonnet 5 | $0.00087 | $0.00087 |
| Haiku 4.5 | $0.00043 | $0.00043 |
Grade A, and why
agentsh 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.
What it actually says
Agent Guidelines
This project cross-compiles to Linux, macOS, and Windows. Follow these rules.
Paths
- Use
filepath.Join()for all path construction - Use
os.TempDir()instead of hardcoded/tmp - Use
filepath.Separatorif you need the separator character - Never hardcode
/or\as path separators
OS-Specific Code
- Use build tags for platform-specific files:
//go:build linux,//go:build darwin,//go:build windows - Check
runtime.GOOSfor runtime platform detection - Skip platform-specific tests with
t.Skip()when features aren't available
Common Platform Differences
| Feature | Linux | macOS | Windows |
|---|---|---|---|
| Unix sockets | Yes | Yes | No |
| PTY | /dev/ptmx | /dev/ptmx | ConPTY |
| Resource limits | cgroups | RLIMIT_* | Job Objects |
| Signals | Full POSIX | Full POSIX | Limited |
| $0 / argv[0] | Works | Works | Different |
Testing
- Run
GOOS=windows go build ./...to verify Windows compilation - Tests using Unix sockets, POSIX signals, or shell features should skip on Windows
- Use
t.TempDir()for test directories (auto-cleaned)
CI-Specific Issues
ConPTY tests hang in CI: Windows ConPTY doesn't properly signal EOF in non-interactive environments. Tests using ConPTY should skip in CI:
func skipInCI(t *testing.T) {
t.Helper()
if os.Getenv("CI") != "" || os.Getenv("GITHUB_ACTIONS") != "" {
t.Skip("skipping ConPTY test in CI")
}
}
Environment Variables
- Use
os.Getenv()/os.Setenv() - On Windows, env vars are case-insensitive but Go preserves case
- Use
os.Environ()to get the full environment asKEY=valueslice
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 First seen · 52 lines · 433 tokens per session scan A a8c26e2efe37
agentsh AGENTS.md is an instructions file published in the GitHub repository canyonroad/agentsh (380 stars, last pushed 27d ago), licensed Apache-2.0. It adds 433 tokens to every session, about $0.0022 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.
Other instructions, from other repositories
hindsight CLAUDE.md
Instructions for vectorize-io/hindsight, covering claude.md, project overview, development commands, local development (api + ui) and start both api server and control plane ui.
agent-framework copilot-instructions.md
Instructions for microsoft/agent-framework, covering github copilot instructions, repository structure and architectural decision records (adrs).
hindsight AGENTS.md
Instructions for vectorize-io/hindsight: See CLAUDE.md for project documentation and coding conventions.
codedb AGENTS.md
AGENTS.md instructions for justrach/codedb, covering codedb agent guidelines, what codedb is (and isn't), review guidelines, pre-merge verification and security-sensitive areas.
agentrq AGENTS.md
Instructions for agentrq/agentrq, covering agentrq codebase notes, project layout, running tests, mcp server (backend/internal/controller/mcp/) and events (experimental).
claude-code-settings copilot-instructions.md
Instructions for feiskyer/claude-code-settings, covering claude.md, environment setup, required dependencies, configuration and skills.