Zero-LLM context routing for AI agents — hooks into every prompt,
injects only the relevant domain, files, and architecture rules.

Any repo, any architecture — docs-first project profiling, no framework-specific detector required.

Tiếng Việt · English

---

AI agents on large codebases default to loading everything — full CLAUDE.md, docs for every domain, models that don't apply. The context window fills; precision drops. The fix isn't a smarter agent. It's a better signal going in.

Open:Context is a Claude Code plugin that fires a UserPromptSubmit hook on every prompt. It tokenizes the task, scores it against your context.yaml, and injects only the matching domain: component chain, relevant files, and applicable architecture rules. If nothing matches, it exits silently. Fully deterministic — no LLM in the routing path.

---

What it looks like

You type a task. Before Claude responds, the hook has already resolved and injected:

[you type]  implement password reset for patron

[injected]  ────────────────────────────────────────────────────────────
            TASK   : implement password reset for patron
            ACTION : create
            ────────────────────────────────────────────────────────────

            [MATCHED DOMAINS]
              member_management   score=2  keywords=['patron']

            [COMPONENTS]
              ▸ CONTROLLER  — instantiates one Operation, renders via Serializer
              ▸ OPERATION   — step_* structure, Form.valid! before any write
              ▸ FORM        — ApplicationForm, validate only, no side-effects
              ▸ MODEL       — AR persistence
              ▸ SERIALIZER  — JSON in Controller, never in Operation

            [RULES]  (4 applicable)
              [CRITICAL] rule-01-no-business-logic-in-controller
              [CRITICAL] rule-02-one-operation-per-action
              [CRITICAL] rule-03-step-method-structure
              [CRITICAL] rule-04-validate-before-mutate

            [FILES]  (3 entries)
              app/controllers/v1/librarians/members_controller.rb
              app/operations/v1/librarians/members/create_operation.rb
              app/models/member.rb

After each matched prompt, Claude Code shows a system notice with token savings:

[open-context] 91% token reduction (1.2 KB injected vs 14.8 KB full context)

Task matches no domain (e.g. "explain this error") → hook exits silently, nothing injected, no notice shown.

---

Install

/plugin marketplace add oopsla5xx/open-context
/plugin install open-context@open-context

Uninstall:

/plugin uninstall open-context@open-context
/plugin marketplace remove open-context

Update to latest version:

claude plugin update open-context@open-context

Run this from your terminal, not as a slash command inside Claude Code — /plugin update doesn't exist. Restart Claude Code afterward to load the new version.

---

Setup

/oc-setup

Asks 3 questions — scope, communication language, and one project-profile confirm — then generates context.yaml and test phrasing files under .open-context/, validates everything in one agentic loop. Re-run any time to reconfigure.

.open-context/ is local to your machine and gitignored (the wizard adds the entry itself) — routing config is per-developer, not shared with the team via git. Each teammate who wants routing runs /oc-setup themselves.

The project-profile question is docs-first: it reads your repo's own README.md/CLAUDE.md/AGENTS.md/docs/**/*.md (found by a deterministic scan, open-context discover-docs) to synthesize language/framework/architecture/actors, citing which file each field came from. No docs? It falls back to reading your source code directly, the way a new engineer would — works for any language or framework, not just the ones with a structured-manifest detector. Stack auto-detect (open-context detect) additionally covers Ruby/Node/Python/Go/Rust/Java manifests as a near-certain cross-check. Details in docs/reference.md.

---

How it works

First run — setup once per project:

flowchart LR
    A[Install plugin] --> B["/oc-setup\n3 questions"]
    B --> C[Generate\ncontext.yaml + tests]
    C --> D[Validate loop\nmax 3 rounds]
    D --> E["✓ Ready"]

Loading

Every prompt — automatic routing:

flowchart LR
    A[User types task] --> B[UserPromptSubmit\nhook]
    B --> C[Tokenize\nScore domains]
    C -->|match| D["Inject\ndomains · files · rules"]
    C -->|no match| E[Silent exit]
    D --> F[Claude responds]
    E --> F

Loading

context.yaml is generated under .open-context/ by /oc-setup or /oc-init — or written by hand anywhere the hook looks (see examples/ for committed reference projects, including one with no layered architecture at all). PyYAML is vendored, so the hook needs no pip install. Full schema in docs/reference.md.

---

Skills

Skill	What it does
/oc-setup	First-run wizard — pre-fills answers from automated discovery, generates context.yaml + tests, validates routing, re-runnable any time
/oc-init	Regenerate context.yaml from existing settings + a scan of docs and source
/oc-resolve <task>	Debug routing — full resolver output, including domains below threshold
/oc-validate	Phrasing coverage + amplification safety check across context.yaml

---

Learn more

• docs/reference.md — automated discovery detail, context.yaml schema, limitations, known unmeasured caveats
• docs/open-context-v0-architecture.md — benchmark methodology
• examples/rails-hmvc-sample/ — working reference project, layered HMVC architecture
• examples/nextjs-sample/ — working reference project, Next.js Server Actions
• examples/data-pipeline-sample/ — working reference project with no architecture.flow — standalone scripts, no layers

---

License

Apache 2.0 — see LICENSE. PyYAML (vendored in vendor/yaml/) is MIT — see vendor/PYYAML_LICENSE.