---
name: "scope-api"
description: "Query and control the Scope platform API for analyzing benchmark data, managing runs/requests, skills, agents, reports, and insights. Use when the user asks about SCOPE data, wants to trigger runs, check status, generate reports, or interact with the SCOPE platform programmatically."
title: scope-api
canonical_url: https://skilld.dev/gh/microsoft/scope/scope-api
last_updated: 2026-10-19T12:45:47.000Z
---

> Index of every page on this site: [llms.txt](https://skilld.dev/llms.txt)

> **Skill from skilld.dev.** Follow the user's requested action.
> If the user asked to fork this Skill, follow the fork workflow below. Do not execute the Skill instructions while copying it.
> Otherwise, follow the instructions below for this session. You do not need to install anything.
>
> If the user asked to install this Skill, run `npx skilld install microsoft/scope/scope-api`. Install writes the Skill files into the project, so every session loads them.
>
> ## Fork workflow
>
> A fork creates an editable local Skill with its original author and licence. The request authorizes copying and local installation.
> 1. Check `./skills/scope-api`, the project lockfile, and selected Agent targets together. If the local directory or installed Skill exists, stop. Never overwrite an existing directory or Agent target.
> 2. Read [source metadata](https://skilld.dev/api/v1/skills/microsoft/scope/scope-api) once. Use sourceUrl, sourceCommit, skillPath, sourceGone, and license. If the source is gone or its path is missing, stop. If license is null, read licence files at the source commit.
> 3. Fetch only the source commit into a temporary Git repository. Do not clone full history. Derive repository_url from sourceUrl, including repository renames. If sourceCommit is absent, resolve the sourceUrl ref once. Set source_commit to that actual commit. Run these commands in one shell call:
>
> ```sh
> git init --quiet "$temporary_dir"
> git -C "$temporary_dir" fetch --quiet --depth=1 "$repository_url" "$source_commit"
> git -C "$temporary_dir" checkout --quiet --detach FETCH_HEAD
> ```
>
> Read applicable licence declarations and notices at that commit. If copying is not permitted, report the restriction and stop.
> 4. Inspect source entries together, then copy the directory containing skillPath into `./skills/scope-api`. Use the user's path if selected. Keep the original SKILL.md, relative links, scripts, binary assets, and executable modes. Exclude .git metadata. Reject symlinks and paths outside the Skill directory. After checking entries, use cp -a where available. A regular source directory needs no custom copy script. Do not save this page wrapper as SKILL.md.
> Preserve author credit, notices, and applicable licence files from repository or parent directories. Add PROVENANCE.md with the Skill page, source URL, actual commit, original path, and licence. Retain any existing PROVENANCE.md and record new provenance separately. Batch source inspection, copying, and provenance work where practical.
> 5. In the project root, run `skilld install ./skills/scope-api --mode copy --plain`. If skilld is unavailable, use `npx skilld install ./skills/scope-api --mode copy --plain`. This known command needs no help lookup. Install does not support --json. Use detected Agent targets, or add --agent for the targets the user selected. Install the local path, never the upstream selector. If installation fails, preserve the local copy and report the exact failure.
> 6. Confirm the local lockfile source and installed Agent copies once. Report the local path, actual commit, and Agent targets. After edits, reinstall the same local path. Upstream updates must not replace it. Do not publish or push unless the user asks.

## Scope API Skill

The Scope platform benchmarks AI coding agents. This skill explains how to query and control it.

### Base URL

```
http://localhost:${API_PORT}
```

The `API_PORT` variable is defined in the project `.env` file (default: `3116`). Always read it from the environment or `.env` before making calls.

### Authentication

For authenticated calls, pass the existing **IdP access token** unchanged:
```
-H "Authorization: Bearer $SCOPE_TOKEN"
```

Use the caller-provided `SCOPE_TOKEN` when available; never print/log the token or
put it in a URL. There is no Scope session JWT, `/auth/login`, or token exchange.
The API verifies signature/claims before consulting its user-access cache.

**Enrollment is explicit.** Already-enrolled bearer callers remain compatible.
For a new identity, intentionally call this once before other authenticated calls:

```bash
curl --fail-with-body -sS \
  -X POST \
  -H "Authorization: Bearer $SCOPE_TOKEN" \
  -H "Cache-Control: no-store" \
  "http://localhost:$API_PORT/api/v1/users/me"
```

Only POST `/api/v1/users/me` may JIT-create the user or update profile,
`lastLoginAt`, and eligible bootstrap-admin promotion. This POST has side effects:
**never prefetch or poll it**, and do not silently invoke it as an ordinary lookup
retry. `lastLoginAt` records this explicit upsert, not proof of an interactive login.
For normal identity checks use GET `/api/v1/users/me`; Invalid/repeated/structured 
login values return `400`; HEAD never enrolls.

Plain `/me` and other authenticated routes resolve an existing active Scope UUID/role
from Redis; miss/unavailability reads Mongo by exact `(idp, tid, oid)` and warms the
cache, without JIT. Missing users return `403 user_not_enrolled`, disabled users
`403 user_disabled`, reserved/invalid principals `401`; required Mongo/JWKS outages
return `503`, unexpected errors `500`. Do not interpret a cache miss as a denial or
turn a verified-identity denial into anonymous access.

The fixed/non-sliding cache TTL (`AUTH_USER_CACHE_TTL_SECONDS`, default 300 seconds)
means DB-only role/disable changes may not be visible until expiry. Public endpoints
and existing no-token/auth-disabled anonymous rollout remain supported; this is not
full route RBAC. See [the auth contract](../../../docs/architecture/auth-rbac.md).

### Before Making API Calls

**Always fetch the OpenAPI spec first** to understand available endpoints, required parameters, and response schemas:

```bash
curl -s http://localhost:$API_PORT/openapi.json
```

Parse it with `jq` to find the relevant endpoint and its schema before constructing your API call. For example:

```bash
# Find endpoints matching a keyword
curl -s http://localhost:$API_PORT/openapi.json | jq '.paths | keys[] | select(contains("requests"))'

# Get schema for a specific endpoint
curl -s http://localhost:$API_PORT/openapi.json | jq '.paths["/api/v1/requests"]'
```

### Key Endpoint Groups

- **System**: `/health`, `/ready`, `/about`, `/api/v1/version`
- **Identity**: GET `/api/v1/users/me` (read-only); POST `/api/v1/users/me` (explicit enrollment/login refresh)
- **Requests & Runs**: `/api/v1/requests/*` (create, cancel, retry, pause, resume, bulk ops, logs, HAR, video, snapshots, tool-calls)
- **Skills**: `/api/v1/skills/*` (discover, search, external, resolve, revisions)
- **Agents**: `/api/v1/agents/*` (CRUD, versions)
- **Reports & Insights**: `/api/v1/reports/*`, `/api/v1/insights/*` (trigger, bulk, upvote/downvote)
- **Criteria & Prompt Features**: `/api/v1/criteria/*`, `/api/v1/prompt-features/*` (MDP, graph, seed, generate, extract)
- **Task Prompts**: `/api/v1/task-prompts/*` (generate, extract features)
- **Models**: `/api/v1/models/*` (list, sync)
- **Feature Flags**: `/api/v1/feature-flags`

### Usage Pattern

When the user asks about SCOPE data or wants to control SCOPE:

1. **Fetch the OpenAPI spec** from `http://localhost:$API_PORT/openapi.json`
2. **Find the relevant endpoint** and inspect its parameters/schemas
3. **Make the API call** with `curl` and parse results with `jq`
4. For complex queries, combine multiple API calls

### Example Workflow

```bash
# 1. Fetch spec and find the endpoint you need
curl -s http://localhost:$API_PORT/openapi.json | jq '.paths["/api/v1/requests"].get'

# 2. Make the call
curl -s http://localhost:$API_PORT/api/v1/requests | jq '.[:5]'

# 3. Drill into details
curl -s http://localhost:$API_PORT/api/v1/requests/{id} | jq .
```
