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 skills add bks-lab/open-bridge --skill a2agit clone --depth 1 https://github.com/bks-lab/open-bridgeWrote 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/skills/bks-lab/open-bridge/a2a)<a href="https://agentmods.dev/skills/bks-lab/open-bridge/a2a"><img src="https://agentmods.dev/badge/skills/bks-lab/open-bridge/a2a/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/skills/bks-lab/open-bridge/a2a"><img src="https://agentmods.dev/badge/skills/bks-lab/open-bridge/a2a.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 | $0.00103 | $0.01147 |
| Opus 5.5 | $0.00041 | $0.00459 |
| Sonnet 5.5 | $0.00021 | $0.00229 |
| Haiku 4.5 | $0.00010 | $0.00115 |
Grade A, and why
a2a 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 5d 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 — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.
a2a
One skill for both directions of A2A: this Bridge calling another agent, and this
Bridge offering an agent of its own. Protocol mechanics only. Who your peers are,
which topics they may ask and in which language lives in data: infra/a2a-peers/,
the agent's own agent.yaml and system-prompt.md, and an org overlay's skill on
top (for example a <org>-a2a skill with the org's peer list and probe prompts).
Engine: skills/a2a/scripts/a2a.py behind skills/a2a/a2a.sh. Stdlib plus PyYAML.
Read the referenced file ONLY when the branch below sends you there.
Commands
| Command | Does |
|---|---|
peers |
list infra/a2a-peers/*.yaml |
card <peer|url> |
fetch the card, report endpoint, version, dialect, https, skills, schemes |
ask <peer> "<question>" [--context ID] [--wait S] [--json] |
send one question in the peer's dialect; --wait polls a held task (a peer waiting for its owner's approval) |
get <peer> <task_id> [--json] |
read a held task again (approved meanwhile?) without sending anything new |
request <peer> "<what>" [--subject S] [--wait S] [--json] |
ask the peer's OWNER to do something; the owner (or a rule the owner set) decides, the model never sees it. Needs owner_request on the card |
rules <peer> [--json] |
the owner's rules for you and your request history at that peer (owner_policy) |
probe <peer> [--prompts FILE] |
no token and wrong token must be refused, then each boundary prompt is checked against its expected refusal or answer |
new-agent <name> --trust public|peer [--port N] [--peer ID:ENV:LOGIN] |
scaffold agents/<name>/; peer writes agent.yaml with auth, a prompt skeleton with a topic boundary, share/ and launch.sh |
Exit codes: 0 clean, 1 a finding or a task that did not complete, 2 usage or a refused input.
The token never passes through you
A peer file names its token as a secret reference (credential_ref: keychain://...). ask and probe re-run themselves under skills/secrets/secrets.sh run --env A2A_TOKEN=<ref>, so the value goes into the child process and nowhere
else. Never ask the user for a token in chat, never put one in a command line.
What ships with it
15 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- a2a.sh 287 B runs code
- assets/peer-agent.yaml.tmpl 1.7 KB
- assets/peer-launch.sh.tmpl 546 B
- assets/peer-system-prompt.md.tmpl 1.1 KB
- assets/probe-prompts.yaml 898 B
- assets/share-README.md 474 B
- references/peer-agent.md 6.2 KB
- references/peers.md 1.2 KB
- references/protocol.md 1.5 KB
- references/public-agent.md 549 B
- references/security-model.md 1.5 KB
- references/traps.md 1.1 KB
- run-tests.sh 170 B runs code
- scripts/a2a.py 22 KB runs code
- tests/test_a2a.py 16 KB runs code
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.
- 5d ago Changed · +6 lines a6a4e47d8868
- 18d ago First seen · 70 lines · 103 tokens per session scan A 96dec0b10536
a2a is a skill published in the GitHub repository bks-lab/open-bridge (13 stars, last pushed today), licensed MIT. It adds 103 tokens to every session and 1,147 once invoked, about $0.0004 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 2026-09-28.
Other skills, from other repositories
agentmail
Use your assigned AgentMail inbox to read email tasks, explicitly send or reply, and check delivery. Provided automatically by your inbox assignment.
mcp-host-styling-integration
Integrates MCP App UI with host theming system. Applies host CSS variables, handles onhostcontextchanged, safe area insets, display mode detection, and fullscreen configuration.
dd-synthesis
Synthesize findings from multiple specialist agents into a coherent investment recommendation. Use when combining reports from 9+ specialist agents into a single decision document.
cog-knowledge-consolidation
Build structured knowledge frameworks from scattered vault notes with source attribution.
formula-authoring
Author TOML-based Formula workflow templates that become Protomolecules and active Molecules in Gas Town's durable workflow system.
incident-postmortem-report
Produce a thorough incident post-mortem report after an outage or customer-impacting event. Covers executive summary, impact, detailed timeline, root cause, contributing factors, corrective and preventive actions, and lessons learned. Use when the user asks to write, draft, or complete a post-mortem, blameless review…