> ## Documentation Index > Fetch the complete documentation index at: https://docs.errorbar.ai/llms.txt > Use this file to discover all available pages before exploring further. # The anchor report > The whole anchor picture for a session, uncached (`Cache-Control: no-store`): the anchor's identity and its live certificate, the rule in force, every stored check, every hold with its state, reason and pinned step (the rule replayed over the history; the current hold carries the reason the trainer was told), the current hold state, and a summary. The same document sits inside the signed environment bundle. From Python: `python -m errorbar_reward.anchor_report [--json]`. Read scope. ## OpenAPI ````yaml /openapi.json get /v1/reward/sessions/{id}/anchor openapi: 3.1.0 info: title: errorbar Management API description: >- The management API behind the improvement loop: capture and setup, request logs and datasets, grades (labels), judges (criteria), evals and deploy gates, fine-tuning and reinforcement learning, dedicated GPU endpoints, and model aliases and versions. Authenticated with a workspace API key (sk_sovereign_...). The inference API (chat, embeddings, rerank, responses) is OpenAI-compatible and documented separately. Responses are snake_case, list endpoints on the loop products use the {"object": "list", "data": [...]} envelope, and refusals use the same nested error shape the gateway emits: {"error": {"message", "type", "code"}}. Request bodies on the loop products (logs, labels, criteria, evals, datasets, aliases) are snake_case; the training and infrastructure products (fine-tuning, GRPO, environment tools, dedicated, model-version adoption) validate camelCase bodies, and each schema below says which it is. Endpoints that spend money require a key minted by a workspace owner or admin and return 403 otherwise. version: 1.0.0 servers: - url: https://gateway.errorbar.ai description: Production - url: https://www.errorbar.ai/api description: Control plane (also served at this base URL) security: - bearerAuth: [] tags: - name: Certified reward description: >- The certified reward for your own trainer: sessions with judge certificates, scoring with a ledger-true kill switch, and the mid-run anchor that holds a run when the reward diverges from an independent judge. paths: /v1/reward/sessions/{id}/anchor: parameters: - name: id in: path required: true schema: type: string get: tags: - Certified reward summary: The anchor report description: >- The whole anchor picture for a session, uncached (`Cache-Control: no-store`): the anchor's identity and its live certificate, the rule in force, every stored check, every hold with its state, reason and pinned step (the rule replayed over the history; the current hold carries the reason the trainer was told), the current hold state, and a summary. The same document sits inside the signed environment bundle. From Python: `python -m errorbar_reward.anchor_report [--json]`. Read scope. operationId: getAnchorReport responses: '200': description: The report. content: application/json: schema: $ref: '#/components/schemas/AnchorReport' '401': $ref: '#/components/responses/Unauthorized' '404': $ref: '#/components/responses/NotFound' '422': description: The session has no anchor. content: application/json: schema: $ref: '#/components/schemas/Error' components: schemas: AnchorReport: type: object properties: session_id: type: string anchor: type: object properties: criterion_id: type: string criterion_name: type: string judge_model: type: - string - 'null' judge_mode: type: string enum: - corrected - observed description: >- corrected — the observed pass rate is mapped through the anchor's measured confusion; observed — no measured error rates, raw pass rate. certificate: description: >- The anchor criterion's full signed certificate as issued now; null if the criterion no longer exists. rule: type: object properties: window: type: integer gap_floor: type: number drop_floor: type: number consecutive: type: integer min_prompts: type: integer description: The thresholds this session decides under. history: type: array items: $ref: '#/components/schemas/AnchorPoint' description: Every stored check, not the session view's 20-point tail. holds: type: array items: $ref: '#/components/schemas/AnchorHoldEvent' description: >- The rule replayed over each prefix of the history; the current hold carries the reason the trainer was told. current: type: object properties: held: type: boolean held_at: type: - string - 'null' held_state: type: - string - 'null' held_reason: type: - string - 'null' pinned_step: type: - string - 'null' resumed_at: type: - string - 'null' summary: type: object properties: checks: type: integer usable_checks: type: integer holds: type: integer max_divergence: type: - number - 'null' max_divergence_step: type: - string - 'null' max_gap_widening: type: - number - 'null' max_gap_widening_step: type: - string - 'null' anchor_range: type: - array - 'null' items: type: number reward_range: type: - array - 'null' items: type: number first_step: type: - string - 'null' last_step: type: - string - 'null' Error: type: object properties: error: type: object properties: message: type: string type: type: string description: >- invalid_request_error, insufficient_quota, rate_limit_error, or api_error. code: type: string description: >- Machine-stable cause, e.g. invalid_api_key, not_found, insufficient_permissions, precondition_failed. required: - message - type - code description: >- Every refusal — gateway and management API alike — uses this one envelope. AnchorPoint: type: object properties: step: type: string at: type: string format: date-time 'n': type: integer description: Prompts both the reward and the anchor judge could grade. reward_rate: type: number description: >- Mean composite grade under the session's reward on the anchor outputs — the signal being optimised. anchor_rate: type: number description: >- The independent anchor judge's pass rate, corrected for its measured error (Lang–Reiczigel) when the criterion carries report counts. anchor_ci: type: array items: type: number minItems: 2 maxItems: 2 anchor_corrected: type: boolean masked: type: integer description: Items the anchor judge could not parse — counted in neither rate. gap_se: type: - number - 'null' description: >- Paired standard error of the gap (reward grade − anchor verdict) over this check's items, stretched by the anchor's 1/(Se+Sp−1). The gap signature tests against 1.96·√(gap_se₀² + gap_seᵢ²), never less than 0.05. Absent on checks stored before it existed. AnchorHoldEvent: type: object properties: step: type: string at: type: string format: date-time state: $ref: '#/components/schemas/AnchorState' reason: type: string pinned_step: type: string AnchorState: type: string enum: - clear - grader_fooled - gamed - degrading description: >- clear — no hold. grader_fooled — the gap (reward read − anchor read on the same outputs) widened beyond its paired noise since the first check while the anchor stayed within its own: the reward passes what the anchor fails, before true quality moves. gamed — the gap widened AND the anchor fell beyond its noise from its best. degrading — the anchor fell and the gap did not widen: both instruments see the fall. Every non-clear state holds. responses: Unauthorized: description: Missing, malformed, or revoked API key. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: message: Invalid API key type: invalid_request_error code: invalid_api_key NotFound: description: The resource does not exist in your workspace. content: application/json: schema: $ref: '#/components/schemas/Error' securitySchemes: bearerAuth: type: http scheme: bearer description: >- Your workspace API key, e.g. `sk_sovereign_...`, sent as `Authorization: Bearer `. ````