Scores

The latest Behavioral Integrity Index and its four signals, and the history.

Method	Path	Scope	Summary
GET	/scores/{agent_id}/history	read	Get Score History
GET	/scores/{agent_id}/latest	read	Get Latest Score
GET	/scores/fleet	read	Get Fleet

Try it

curl -sf -X POST "$VIGIL_URL/agents/" -H "Authorization: Bearer $VIGIL_KEY" -H "Content-Type: application/json" \
  -d '{"agent_id":"docs-api","name":"Docs API"}' > /dev/null || true
curl -sf -X POST "$VIGIL_URL/events/" -H "Authorization: Bearer $VIGIL_KEY" -H "Content-Type: application/json" \
  -d '{"agent_id":"docs-api","event_type":"output_generated","payload":{"content_length":512},"dti":0.9}' > /dev/null
curl -sf -H "Authorization: Bearer $VIGIL_KEY" "$VIGIL_URL/scores/docs-api/latest"

Routes

GET /scores/{agent_id}/history

Get Score History. Scope: read.

Parameter	In	Required	Type	Notes
agent_id	path	yes	string
limit	query	no	integer

Response 200:

[
  {
    "agent_id": "string",
    "bii": 0,
    "score_inflation_signal": 0,
    "config_tamper_signal": 0,
    "alignment_faking_signal": 0,
    "asset_movement_signal": 0,
    "blueprint_drift_signal": 0,
    "alignment_p": 0,
    "alignment_cliffs_delta": 0,
    "composite_trust": 0,
    "event_window": 0,
    "computed_at": "string",
    "policy_digest": "string"
  }
]

GET /scores/{agent_id}/latest

Get Latest Score. Scope: read.

Parameter	In	Required	Type	Notes
agent_id	path	yes	string
dti	query	no	number | null

Response 200:

{
  "agent_id": "string",
  "bii": 0,
  "score_inflation_signal": 0,
  "config_tamper_signal": 0,
  "alignment_faking_signal": 0,
  "asset_movement_signal": 0,
  "blueprint_drift_signal": 0,
  "alignment_p": 0,
  "alignment_cliffs_delta": 0,
  "composite_trust": 0,
  "event_window": 0,
  "computed_at": "string",
  "policy_digest": "string"
}

GET /scores/fleet

Get Fleet. Scope: read.

The latest score per agent in the org. With adjust=bh the alignment faking p-values pass through Benjamini-Hochberg at the policy's detectors.fdr.q, so a large fleet does not page on chance; alignment_significant is then p_adjusted <= q. Without it, alignment_significant is the raw p < alpha and alignment_p_adjusted is null. Agents whose latest score had no test (an arm below min_n) have null p and are never significant.

Parameter	In	Required	Type	Notes
adjust	query	no	bh | null	bh: Benjamini-Hochberg over the alignment p-values at the policy's fdr.q

Response 200:

{
  "adjust": "string | null",
  "method": "string | null",
  "q": 0,
  "alpha": 0,
  "n_tested": 0,
  "n_significant": 0,
  "agents": [
    {
      "agent_id": "string",
      "bii": 0,
      "score_inflation_signal": 0,
      "config_tamper_signal": 0,
      "alignment_faking_signal": 0,
      "asset_movement_signal": 0,
      "blueprint_drift_signal": 0,
      "alignment_p": 0,
      "alignment_cliffs_delta": 0,
      "composite_trust": 0,
      "event_window": 0,
      "computed_at": "string",
      "policy_digest": "string",
      "alignment_p_adjusted": 0,
      "alignment_significant": true
    }
  ]
}

Generated from openapi.json (VIGIL 0.2.0). Scopes follow the router dependencies in VIGIL_SPEC.md section 4.2.