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/oxbshw/agent-span/agents-mdgit clone --depth 1 https://github.com/oxbshw/Agent-SpanWhat 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 | $0.00700 | $0.00700 |
| Opus 5 | $0.00350 | $0.00350 |
| Sonnet 5 | $0.00140 | $0.00140 |
| Haiku 4.5 | $0.00070 | $0.00070 |
Grade A, and why
Agent-Span 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.
How it starts
The opening of the file, as written. The whole thing — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
AgentSpan is an async-Rust web access gateway for AI agents. This file is for AI coding assistants (Claude Code, Cursor, Codex) working in this repo.
Build & test
cargo test --workspace # run all tests
cargo clippy --workspace --all-targets -- -D warnings # lint
cargo fmt --all -- --check # format check
cargo run --bin agentspan -- serve # start the API on :8080
cargo run --bin agentspan-mcp # start the MCP server (stdio)
Python SDK: pip install -e sdk/python[dev] then pytest -q in sdk/python/.
Crate layout
| Crate | What lives here |
|---|---|
agentspan-core |
Channel/Backend traits, types, config, errors |
agentspan-probe |
Command execution + 6-state health classification |
agentspan-router |
Backend selection, retry, circuit breaker, adaptive routing |
agentspan-cache |
3-tier cache (memory/disk/Redis), singleflight, optimizer |
agentspan-auth |
API keys (SHA-256), tenants, RBAC, rate limiting, audit |
agentspan-channels |
All 52 channel implementations + registry |
agentspan-api |
Axum REST server, SSE, OpenAPI spec, middleware |
agentspan-mcp |
MCP server (stdio + HTTP), 91 tools |
agentspan-cli |
The agentspan binary — install, doctor, serve, skill, etc. |
Adding a channel
- Copy
crates/agentspan-channels/src/hackernews.rs(free API) orspotify.rs(OAuth) as a template. - Implement
Channel+Backendtraits. UseBackendRouterfor fallback. - Register in
src/lib.rs(module + re-export) andsrc/registry.rs. - Add MCP tools in
crates/agentspan-mcp/src/tools.rs. - Add a format rule in
crates/agentspan-cli/src/commands/format.rs. - Run
cargo test -p agentspan-channels— the registry count test will tell you the expected count.
Adding an MCP tool
Add a ToolDef entry in crates/agentspan-mcp/src/tools.rs. The channel
field must match a registered channel name. Use Op::Read for URL-based
fetch, Op::Search for keyword queries, Op::Doctor for health.
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 · 61 lines · 700 tokens per session scan A fb32326c26d7
Agent-Span AGENTS.md is an instructions file published in the GitHub repository oxbshw/Agent-Span (42 stars, last pushed 1mo ago), licensed MIT. It adds 700 tokens to every session, about $0.0035 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
substack-gateway-oss CLAUDE.md
Instructions for jakub-k-slys/substack-gateway-oss, covering claude.md, commands, install dependencies (dev included, all workspace members), run the server (dev mode with reload) and lint.
fieldflow AGENTS.md
Instructions for guillaumegay13/fieldflow, covering repository guidelines, project structure & module organization, current product scope, build, test, and development commands and coding style & naming conventions.
tray-api-ai-plugin AGENTS.md
Instructions for tray-tecnologia/tray-api-ai-plugin, covering tray api plugin, regras obrigatórias para toda sessão, autenticação, url e formato and rate limit.
tray-api-ai-plugin copilot-instructions.md
Instructions for tray-tecnologia/tray-api-ai-plugin, covering tray api plugin, regras obrigatórias para toda integração tray, autenticação oauth 2.0, formato de requisições and rate limit.
tray-api-ai-plugin CLAUDE.md
Instructions for tray-tecnologia/tray-api-ai-plugin, covering tray api plugin — contexto para claude, regras obrigatórias, bloco mandatory e lint de skills, validação local and busca em docs.
substack-gateway-oss AGENTS.md
Instructions for jakub-k-slys/substack-gateway-oss, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.