opchain

io.github.asfbay-bitPublished by GitHub account asfbay-bitv2.1.0Updated Nov 6, 2026

Claude Code / MCP skills for the dev pipeline: discover, spec, design, build, ship, operate.

VerifiedStreamable HTTPWeb executableKnowledge & MemoryDeveloper Tools

Overview

AI-generated overview

Remote MCP server that serves opchain's software development pipeline skills and lets an assistant route, load and checkpoint them.

What you can do
  • List and route skills: enumerate all opchain skills and map a request or /oc-* command to the right skill and entry phase
  • Load skill instructions: fetch the shared orchestrator protocol and the full SKILL.md text for a chosen skill to follow
  • Manage checkpoints: create a private server-issued session, then read and write a skill's saved progress so work resumes across sessions
When to use it
Useful when you want an assistant to follow opchain's structured discover, spec, design, build, audit, ship workflow for software projects, and to carry progress between sessions through checkpoints.
Requirements
A remote MCP endpoint at opchain.dev over streamable HTTP; no authentication, environment variables or headers are declared. The README's local skill installs and runtime commands are separate from this endpoint and may need Node.js 22.13 or newer.
Before you install
The server returns instruction text that assistants are told to follow verbatim, so review what a skill asks for before letting it act. Checkpoints are stored server-side under a session token that should not be shared. Skills cover writing, committing, deploying and auditing code, which can change repositories and production systems.

Installation

In SourceWeft

  1. Open opchain in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/io.github.asfbay-bit%2Fopchain-skills/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config.

{
  "mcpServers": {
    "opchain": {
      "type": "http",
      "url": "https://opchain.dev/mcp"
    }
  }
}

README

opchain

Opchain 2.1.0 · 36 skills · released November 5, 2026.

A coordinated set of Claude skills covering the full software development pipeline — discover, spec, design, build, audit, ship, scale. One skill per phase. A shared JSON checkpoint protocol carries context across sessions and skills, so work resumes where you left off.

Site & docs: https://opchain.dev

install

Claude Code — plugin (recommended)

Ships complete skills and plugin hooks for session context and suggestions. Run /oc-enroll in each repository to enable the Git commit gate. Installing the plugin alone does not enroll repositories or enable learning.

/plugin marketplace add asfbay-bit/opchain-skills/plugin install opchain

Claude Code — zip (skills only, no hooks)

bash
curl -L https://opchain.dev/opchain-skills.zip -o opchain-skills.zipunzip opchain-skills.zip -d .claude/skills/claude> /oc-discover

Claude.ai / Claude Desktop

Download complete skill ZIPs from Install, then use your host's supported skill upload flow. Keep references and runtime files with each skill. Local runtime commands require Node.js 22.13 or newer and a host that can execute them; a hosted chat without local tools cannot run those commands.

Team (check into git)

bash
git add .claude/skills/git commit -m "chore: add opchain skills"git push

Upgrade

For an existing repo installation, ask your coding agent to run:

text
/oc-update check/oc-update

If you use 1.9 or earlier and the updater is missing, copy the upgrade prompt at the top of Install. It checks the host and published release, then installs the complete package. The updater preserves checkpoints, local tracking consent, usage history and unrelated skills, and backs up replaced Opchain files. Global plugin installations use the host's supported plugin update.

Open a new agent session afterward and run /oc-update check to verify the installed package. Learning remains off until explicitly enabled; provider access and an external reviewer must be configured separately.

skills

skillwhat it does
oc-app-architectDiscover → spec → build, with Generator/Evaluator QA loop
oc-stack-forgeStack advisor across Cloudflare, Vercel, AWS, Supabase, etc.
oc-ux-engineerDesign Planner → Generator → Evaluator harness
oc-dash-forgeDashboards and dense-data UIs (spec + React prototype)
oc-integrations-engineerThird-party API integrations (Slack, Stripe, OAuth)
oc-api-devFirst-party API design (OpenAPI, GraphQL, SDKs)
oc-claude-apiClaude API apps: model routing, prompt caching, tool use
oc-rag-forgeRAG systems: vector DBs, embeddings, chunking, retrieval eval
oc-agent-forgeClaude Agent SDK apps: topology, tool budgets, harness loops
oc-prompt-opsPrompts as code: versioning, eval datasets, regression detection
oc-qa-opsTest-pyramid design: coverage budgets, contract-test matrix, load plans
oc-data-opsData pipelines: ingestion, dbt layering, observable data contracts
oc-signal-forgeQuestion → trustworthy metric: instrument, harvest, verify, wire
oc-code-auditorAuditor → Fixer → Verifier; 5-layer pre-deploy sweep
oc-security-auditorThreat modeling, OWASP hardening, attack-surface review
oc-security-hardeningExecutes the fixes + per-deploy hardening gate (/oc-harden)
oc-compliance-opsControl register + audit-ready evidence bundles at deploy time
oc-bug-checkPre-commit QA gate: types, lint, tests, anti-patterns, secrets
oc-docs-forgePR documentation packets, README/changelog/ADR upkeep
oc-repo-opsRepo hygiene + PR readiness gate: catalogs, generated files
oc-git-opsBranches, commits, PRs, sync
oc-deploy-opsAudit gate → staging → production with rollback
oc-fleet-opsSelf-managed fleets: IaC, multi-container rollouts, day-2 ops
oc-modularize-opsLive-monolith decomposition with golden-fixture proof
oc-monitoring-opsPost-deploy observability — uptime, errors, alerts
oc-scale-opsLoad tests, perf budgets, caching, capacity planning
oc-migration-opsDB / framework / auth / platform migrations
oc-cost-opsLLM cost attribution, budget gates, model-tier routing
oc-telemetry-opsOpt-in local usage metering, anonymized dashboard aggregates
oc-reverse-specReverse-engineer existing code into spec docs
oc-release-opsVersion bumps, changelogs, release announcements
oc-orchestratorCross-skill status and routing (/oc-ops)
oc-checkpoint-protocolShared session-persistence schema (bundled in every skill)
oc-updateUpdates complete skill and runtime files while preserving local state
oc-hindsightSource-backed lessons, retrieval evaluation and external review
oc-evolveTask-tested rule proposals with regression checks and external approval

Full descriptions, triggers, and examples: https://opchain.dev/skills

the checkpoint protocol

Every skill writes a JSON checkpoint to .checkpoints/ in your project. Skills read each other's checkpoints to make informed decisions — context flows forward without manual handoffs.

  • oc-deploy-ops reads oc-code-auditor — CRITICAL findings block deploy
  • The build evaluator reads oc-ux-engineer — grades frontend against the approved design spec
  • oc-git-ops reads oc-app-architect — names branches by sprint

Schema lives in skills/oc-checkpoint-protocol/SKILL.md.

this repository

This is a public, force-push snapshot mirror of the skill source. The site, build system, and internal tooling live in a private upstream repo; only the product (this directory) is published here, refreshed on every change to main upstream.

See CONTRIBUTING.md for how issues and PRs flow back into the upstream repo.

license

Apache-2.0 — see LICENSE and NOTICE. Releases up to and including v1.8.2 were published under MIT.

Source: README.md at commit 1c5878a

Tools

7

list_skills

List every opchain skill (id, displayName, description, phases, commands). Call this first to see the concept→ship pipeline.

Takes no parameters.

route

Map a request or an /oc-* command to the skill that should handle it, with the entry phase. Use when you're unsure which skill fits.

Parametersqueryrequired

get_skill

Return the full SKILL.md (the instructions) for one skill id. Load it, then follow it. Read get_orchestrator once before your first get_skill in a session.

Parametersidrequired

get_orchestrator

Return the shared orchestrator protocol (welcome flow, pipeline map, active-chaining rules). Read once at the start of a session.

Takes no parameters.

create_checkpoint_session

Create a private server-issued session for checkpoint storage. Retain the returned sessionId and use it with read_checkpoint/write_checkpoint; never invent or share one.

Takes no parameters.

read_checkpoint

Read a skill's saved session checkpoint (where you left off). sessionId must be a private token returned by create_checkpoint_session.

ParametersskillrequiredsessionIdrequired

write_checkpoint

Persist a skill's session checkpoint so progress survives across sessions. Stores the object as given (any JSON object up to 64 KiB); it does not validate it against the opchain checkpoint protocol, so write the shape the protocol documents.

ParametersskillrequiredsessionIdrequiredcheckpointrequiredexpectedRevision

Version history

4
  1. v2.1.0LatestNov 6, 2026
  2. v2.0.4Sep 27, 2026
  3. v2.0.3Sep 19, 2026
  4. v2.0.2Sep 16, 2026

Score

100% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listedRequired15 pts
  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts