feat(webhooks): outbound webhook delivery

[RustMunkey]

Summary by CodeRabbit

• New Features

  • Multi-model selection with tier-based access and optional model override plus per-run system prompt.
  • Per-model billing multipliers applied to usage accounting.
  • Outbound webhooks: CRUD, test deliveries, delivery logs, and background dispatch with retries.

• Chores

  • CI and test suite updated to include the runtime service.
  • Lint-staged configuration replaced with an environment-aware implementation.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Free

Run ID: f3fc3ba5-7268-4350-ad3c-1b980e957b0d

📥 Commits

Reviewing files that changed from the base of the PR and between 88fbee9 and 255ee23.

⛔ Files ignored due to path filters (1)

• pnpm-lock.yaml is excluded by !**/pnpm-lock.yaml

📒 Files selected for processing (18)

• .github/workflows/codeql.yml
• packages/db/src/schema/pg/enums.ts
• packages/db/src/schema/pg/index.ts
• packages/db/src/schema/pg/nodes.ts
• packages/db/src/schema/pg/relations.ts
• packages/jobs/src/dispatch.ts
• packages/webhooks/src/deliver.ts
• packages/webhooks/src/events.ts
• packages/webhooks/src/index.ts
• packages/webhooks/src/sign.ts
• services/api/package.json
• services/api/src/index.ts
• services/api/src/jobs/webhook.ts
• services/api/src/routes/webhooks.ts
• services/worker/src/worker/consumer.py
• services/worker/src/worker/handlers/__init__.py
• services/worker/src/worker/handlers/webhook.py
• services/worker/src/worker/models.py

---

📝 Walkthrough

Walkthrough

Adds a new @maschina/model package with tiered model catalog and billing multipliers; integrates model selection into agent dispatch, extends job/daemon/runtime payloads for model and system prompt, implements runtime routing between Ollama and cloud providers with token multipliers, and introduces webhook delivery infrastructure and node DB schemas.

Changes

Cohort / File(s)	Summary
Model package packages/model/package.json, packages/model/tsconfig.json, packages/model/src/catalog.ts, packages/model/src/catalog.test.ts, packages/model/src/index.ts	New package providing MODEL_CATALOG, tier defaults, access validation, multiplier lookup, resolver, and tests.
Runtime & tests services/runtime/src/runner.py, services/runtime/tests/test_runner_routing.py	Runtime routing by model prefix (ollama/* → local; others → Anthropic), billing multiplier computation, token accounting, and unit tests for routing/multiplier helpers.
Daemon / orchestrator services/daemon/src/runtime/mod.rs, services/daemon/src/orchestrator/scan.rs, services/daemon/src/orchestrator/scan_compat.rs, services/daemon/src/orchestrator/analyze.rs	Expanded job/run structs to include model and system_prompt; renamed RunOutput.payload → output_payload; runtime request/endpoint updated to include plan_tier, model, system_prompt, max_tokens, timeout_secs and POST /run; SQL binding fixed to output_payload.
API layer & routes services/api/src/routes/agents.ts, services/api/src/routes/webhooks.ts, services/api/src/index.ts, services/api/src/jobs/webhook.ts, services/api/Dockerfile, services/api/package.json	Validate/resolve model per tier, compute system prompt, persist model & systemPrompt, dispatch jobs with new fields; new webhook CRUD, test endpoint, delivery listing, Stripe inbound handling, and started webhook dispatcher on service startup; added @maschina/model dependency and Docker build ordering.
Jobs & types packages/jobs/src/types.ts, packages/jobs/src/dispatch.ts	AgentExecuteJob and dispatchAgentRun extended with model and systemPrompt (and timeoutSecs); added dispatchWebhookJob for worker webhook dispatch subject.
Validation schema packages/validation/src/schemas/agent.ts	RunAgentSchema gains optional model string field for model override.
Worker webhook handler & models services/worker/src/worker/handlers/webhook.py, services/worker/src/worker/models.py, services/worker/src/worker/consumer.py, services/worker/src/worker/handlers/__init__.py	New WebhookDispatchJob model and handler implementing signed POSTs, retries/backoff, delivery state updates; consumer routes webhook_dispatch subject to handler; handlers export updated.
Webhooks package (types, signing, delivery) packages/webhooks/src/events.ts, packages/webhooks/src/sign.ts, packages/webhooks/src/deliver.ts, packages/webhooks/src/index.ts	Typed webhook event payloads, signing utilities, delivery logic with backoff and per-attempt timeouts, and consolidated exports.
DB schema: nodes packages/db/src/schema/pg/nodes.ts, packages/db/src/schema/pg/enums.ts, packages/db/src/schema/pg/relations.ts, packages/db/src/schema/pg/index.ts	New nodes, node_capabilities, node_heartbeats tables; new enums for node status/tier; relation mappings and index export added.
CI / tooling / scripts .github/workflows/ci.yml, .github/workflows/codeql.yml, .lintstagedrc.json, .lintstagedrc.mjs, package.json	Added model to build/test flows; replaced JSON lint-staged with .mjs that filters existing files; added pytest runtime and other pytest scripts; CodeQL upgraded to v4 with continue-on-error.
CHANGELOG CHANGELOG.md	Documented model routing, API/daemon/runtime changes, added tests and CI updates.

sequenceDiagram
    participant Client
    participant API
    participant Validation as ModelValidation
    participant Jobs
    participant Daemon
    participant Runtime

    Client->>API: POST /agents/run (agentId, tier, model?)
    API->>Validation: validateModelAccess(tier, model)
    alt denied
        Validation-->>API: 403 reason
        API-->>Client: 403
    else allowed
        Validation-->>API: allowed
        API->>Validation: resolveModel(tier, model)
        Validation-->>API: resolvedModel
        API->>API: systemPrompt = agent.config or default
        API->>Jobs: dispatchAgentRun(runId, model, systemPrompt, timeoutSecs, ...)
        Jobs-->>API: jobAccepted
        API-->>Client: {runId, status}
        Jobs->>Daemon: agent.execute job (model, system_prompt, plan_tier, timeout_secs, ...)
        Daemon->>Runtime: POST /run (model, system_prompt, plan_tier, max_tokens, timeout_secs, input_payload)
        alt model starts with "ollama/"
            Runtime->>Runtime: route to OllamaRunner (multiplier: 0)
        else cloud model
            Runtime->>Runtime: lazy-import Anthropic client (multiplier: 1|3|15)
        end
        Runtime-->>Daemon: RunResponse (output_payload, raw_tokens, billed_tokens)
        Daemon->>Daemon: persist output_payload (RunOutput.output_payload)
        Daemon-->>Jobs: mark complete
    end

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~60 minutes

🐰 I hopped through code with a tiny thump,
Catalogs, routes, and multipliers jump—
Ollama or cloud, the runtime now knows,
Webhooks delivered where the wild wind blows,
Tokens counted, prompts carried, the warren hums.

📝 Coding Plan

• Generate coding plan for human review comments

---

Note

🎁 Summarized by CodeRabbit Free

Your organization is on the Free plan. CodeRabbit will generate a high-level summary and a walkthrough for each pull request. For a comprehensive line-by-line review, please upgrade your subscription to CodeRabbit Pro by visiting https://app.coderabbit.ai/login.

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

Tip

CodeRabbit can enforce grammar and style rules using `languagetool`.

Configure the reviews.tools.languagetool setting to enable/disable rules and categories. Refer to the LanguageTool Community to learn more.