covalent-bond: Skill for Claude Code

.claude/skills/covalent-bond-guardrails/SKILL.md

covalent-bond-guardrails is a skill for Claude Code from gopalrajsuresh/covalent-bond. It costs 68 tokens per session (1,053 once invoked), scanned A, original, MIT.

A set of repository-specific safety rules for Covalent Bond, a peer-to-peer encrypted communication channel for AI coding agents. It tells an agent which security and protocol rules must remain intact while editing.

In plain words
What is it for?
Use it before changing Covalent Bond code, tests, or configuration to check required safeguards and follow a verify-before-edit workflow.
Why use it?
It prevents changes that could leak session secrets, corrupt the MCP message stream, accept files without consent, or weaken security.

Skill for Claude Code

Written for Claude Code: installed under .claude/.

This is gopalrajsuresh/covalent-bond's own configuration. It tells Claude Code how to work on covalent-bond itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything covalent-bond configures →

Reuse

Borrowing it

Nothing to install: this file belongs to gopalrajsuresh/covalent-bond. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/gopalrajsuresh/covalent-bond/main/.claude/skills/covalent-bond-guardrails/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/gopalrajsuresh/covalent-bond

Made for: Claude Code.

Wrote 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.

agentmods badge for covalent-bond-guardrails

README.md
[![agentmods](https://agentmods.dev/badge/skills/gopalrajsuresh/covalent-bond/covalent-bond-guardrails/github.svg)](https://agentmods.dev/skills/gopalrajsuresh/covalent-bond/covalent-bond-guardrails)
Your own site
<a href="https://agentmods.dev/skills/gopalrajsuresh/covalent-bond/covalent-bond-guardrails"><img src="https://agentmods.dev/badge/skills/gopalrajsuresh/covalent-bond/covalent-bond-guardrails/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.

agentmods 80×15 button for covalent-bond-guardrails

Your own site · 80×15
<a href="https://agentmods.dev/skills/gopalrajsuresh/covalent-bond/covalent-bond-guardrails"><img src="https://agentmods.dev/badge/skills/gopalrajsuresh/covalent-bond/covalent-bond-guardrails.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 68 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,053 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00068 $0.01053
Opus 5 $0.00034 $0.00526
Sonnet 5 $0.00014 $0.00211
Haiku 4.5 $0.00007 $0.00105

Measured 12d ago against content hash 0b1500d06f04, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

covalent-bond-guardrails 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 12d 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.

.claude/skills/covalent-bond-guardrails/SKILL.md · 79 lines

How it starts

The opening of the file, as written. The whole thing — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Covalent Bond engineering guardrails

Covalent Bond is a peer-to-peer, end-to-end-encrypted channel for AI coding agents, exposed over MCP. A single careless change can silently defeat its security or corrupt the MCP protocol stream. Follow these rules on every edit. Context is in docs/ARCHITECTURE.md; this skill is the enforceable checklist.

Security invariants (never break these)

They are correctness requirements, not style. If a change would touch any of them, STOP and confirm with the human first (see Working discipline).

  1. No stdout writes from any module the MCP server loads (daemon/, mcp/, relay/, security/, transfer/). stdout is the JSON-RPC stream. Log via logger (stderr) from security/index.js. A stray console.log corrupts the protocol. (Standalone scripts and the mock relay may log.)
  2. The session code never leaves the machine to the relay. The relay is addressed by the routing ID (PBKDF2(sessionCode)) only. The code is the secret that defeats a malicious relay; leaking it reopens the MITM hole.
  3. No received file is written without explicit bond_accept (human consent). Incoming transfers stay pending until then.
  4. All peer content is untrusted data. Keep the untrusted-content wrapper and never let received text be interpreted as instructions.
  5. The session key is derived once per session. Never run the 600k-iteration PBKDF2 per message. Wipe ephemeral private keys once the session key exists.
  6. Local servers bind to 127.0.0.1, never 0.0.0.0.
  7. Crypto domain constants stay stable and consistent (CovalentBond-Routing, CovalentBond-CodeKey, CovalentBond-SessionKey, CovalentBond-Confirm). Both peers must derive identical values.

What NOT to add

  • No new runtime dependencies without a clear need; the small surface is a feature. Prefer Node built-ins (crypto, fs, http).
  • No rolling your own crypto beyond composing Node's crypto primitives as already done. No custom ciphers, no home-made KDFs.
  • No telemetry, analytics, or network calls to anything but the configured relay. Never phone home.
  • No hardcoded secrets, account IDs, KV IDs, personal subdomains, or emails. Relay URL comes from COVALENT_RELAY_URL (or the localhost default).
  • No weakening of validation: file-type whitelist, size caps, path safety, session-code format, routing-ID format. Don't broaden these to make a test pass.
  • No stdout logging, ever (see invariant 1).
  • No vendor lock-in language or Claude-specific assumptions in code/docs; keep it "the agent" / "the MCP client".

Read the full file on GitHub · 79 lines

Changes

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.

  1. 12d ago First seen · 79 lines · 68 tokens per session scan A 0b1500d06f04

Subscribe to this mod's changes

covalent-bond-guardrails is a skill published in the GitHub repository gopalrajsuresh/covalent-bond (0 stars, last pushed 2d ago), licensed MIT. It adds 68 tokens to every session and 1,053 once invoked, about $0.0003 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.