Introduction
Connect an agent or an app to BDC Schedules of Investments: the fund tape, the credit, and every holder mark that was filed.
Data is taken from SEC 10-Q and 10-K filings, plus opt-in N-CSR / N-CSRS interval funds. Each line traces to the filing it came from.
What you can access
- Find a credit - search a borrower the way you do in the terminal.
- Holder marks - every fund that printed that name, and the mark they filed.
- Fund profile - Ares, Blackstone, Blue Owl, and the rest, by CIK or ticker.
- SOI tape - the fund's Schedule of Investments, line by line (company, investment, cost, fair value, mark).
- Non-accrual - lines the fund footnoted as non-accrual.
- Coverage - how many funds and credits are in the latest snapshot.
Coverage
| Funds | Credits | Fair value | Source |
|---|---|---|---|
| Live counts: filers, credits, and fair value on GET /v1/coverage. BDCs file 10-Q / 10-K. Interval funds on the roster file N-CSR / N-CSRS. | |||
Live counts: GET /v1/coverage.
Data notes
- Structured JSON - the same field names on every endpoint.
- As-reported names -
company_rawandinvestment_raware the cells as printed on the SOI. - Marks - fair value / cost from the filed schedule.
1.00is par. - Money - $ millions.
- Funds - identified by CIK. Ticker is extra when the fund is listed.
Built for
- Credit agents and LLM pipelines
- Private-credit research
- Portfolio and risk systems
- Internal data layers
Start with the Quick start.
Quick start
Authenticate and make your first request.
1. Get an API key
Get a key at API plans. Trial on a work email, or subscribe to Tape or Matched. You see the secret once. Copy it. Terminal seats are by invitation.
2. Make your first request
Every request needs your key in the X-API-KEY header.
curl "https://dirigon.com/v1/filers?q=ARCC" \ -H "X-API-KEY: YOUR_KEY"
import requests
headers = {"X-API-KEY": "YOUR_KEY"}
r = requests.get("https://dirigon.com/v1/filers?q=ARCC", headers=headers)
fund = r.json()["filers"][0]
print(fund["name"], fund["cik"], fund["total_fv_m"])
3. Open a tape, or find a credit
curl "https://dirigon.com/v1/filers/1287750/positions?limit=5" \ -H "X-API-KEY: YOUR_KEY" curl "https://dirigon.com/v1/issuers?q=Coupa" \ -H "X-API-KEY: YOUR_KEY"
import requests
headers = {"X-API-KEY": "YOUR_KEY"}
tape = requests.get(
"https://dirigon.com/v1/filers/1287750/positions?limit=5",
headers=headers,
).json()
credit = requests.get(
"https://dirigon.com/v1/issuers?q=Coupa",
headers=headers,
).json()["issuers"][0]
print(credit["issuer_id"], credit["display_name"])
What's next
Find a credit
Search a borrower. Same idea as Find credit in the terminal.
Look up a name, get an issuer_id, then open it. Use that id for holder marks.
GET/v1/issuers?q=Coupa
GET/v1/issuers/ISS-002537
Query
| Param | Where | Meaning |
|---|---|---|
| q | query | Name search. |
| issuer_id | path | Credit id (ISS-######). |
| limit | query | How many rows to return. Default 100, max 500. |
| cursor | query | From next_page_url. |
Example
curl "https://dirigon.com/v1/issuers?q=Coupa" \ -H "X-API-KEY: YOUR_KEY"
import requests
headers = {"X-API-KEY": "YOUR_KEY"}
r = requests.get("https://dirigon.com/v1/issuers?q=Coupa", headers=headers)
credit = r.json()["issuers"][0]
print(credit["issuer_id"], credit["display_name"], credit["n_holders"])
Response
{
"issuers": [
{
"issuer_id": "ISS-002537",
"display_name": "Denali Midco 2",
"blocked": false,
"n_holders": 32,
"total_fv_m": 2785.9
}
]
}
Fields
| Field | Meaning |
|---|---|
| issuer_id | Credit id. Use this in later calls. |
| display_name | Credit name. |
| n_holders | How many funds print it. |
| total_fv_m | Fair value across those holders, $ millions. |
| blocked | True if the name is disputed. Skip it for comparisons. |
Holder marks
Every fund that printed the credit, and the mark they filed.
This is the disagreement view. One row per holder. Mark is fair value / cost.
GET/v1/issuers/ISS-002537/marks
If blocked is true on the credit, this endpoint returns 404. Skip disputed names for comparisons.
Example
curl "https://dirigon.com/v1/issuers/ISS-002537/marks" \ -H "X-API-KEY: YOUR_KEY"
import requests
headers = {"X-API-KEY": "YOUR_KEY"}
r = requests.get("https://dirigon.com/v1/issuers/ISS-002537/marks", headers=headers)
for row in r.json()["marks"]:
print(row["filer"], row["mark"], row["fv_m"])
Response
{
"marks": [
{
"issuer_id": "ISS-002537",
"cik": 1287750,
"filer": "ARES CAPITAL CORP",
"mark": 0.9524,
"fv_m": 365.8,
"cost_m": 384.1,
"period_end": "2026-06-30"
}
]
}
Fields
| Field | Meaning |
|---|---|
| filer | Fund name. |
| cik | Fund CIK. |
| mark | Fair value / cost (1.00 = par). |
| fv_m | Fair value, $ millions. |
| cost_m | Amortized cost, $ millions. |
| period_end | SOI period. |
Fund profile
Same idea as Fund profile in the terminal. Search by ticker, name, or CIK.
GET/v1/filers?q=ARCC
GET/v1/filers/1287750
Query
| Param | Where | Meaning |
|---|---|---|
| q | query | Name, ticker, or CIK. |
| cik | path | SEC CIK of the fund. |
| limit | query | How many rows to return. Default 100, max 500. |
Example
curl "https://dirigon.com/v1/filers?q=ARCC" \ -H "X-API-KEY: YOUR_KEY"
import requests
headers = {"X-API-KEY": "YOUR_KEY"}
r = requests.get("https://dirigon.com/v1/filers?q=ARCC", headers=headers)
fund = r.json()["filers"][0]
print(fund["name"], fund["cik"], fund["n_positions"], fund["total_fv_m"])
Response
{
"filers": [
{
"cik": 1287750,
"name": "ARES CAPITAL CORP",
"ticker": "ARCC",
"manager": "Ares Management",
"period_end": "2026-06-30",
"n_positions": 1239,
"total_fv_m": 28338.3,
"form": "10-Q",
"filing_url": "https://www.sec.gov/Archives/edgar/data/1287750/000162828026050307/arcc-20260630.htm"
}
]
}
Fields
| Field | Meaning |
|---|---|
| cik | SEC id for the fund. |
| name | Fund name. |
| ticker | Listed ticker when there is one. |
| manager | Adviser. |
| period_end | Latest SOI period. |
| n_positions | Lines on the tape. |
| total_fv_m | Fair value of that tape, $ millions. |
| filing_url | Link to the SEC filing. |
SOI tape
The Schedule of Investments for one fund, line by line: company, investment, cost, fair value, mark.
GET/v1/filers/1287750/positions?limit=25
limit is how many rows you want in total. Follow next_page_url when it is present.
Query
| Param | Where | Meaning |
|---|---|---|
| cik | path | Fund CIK (required). |
| limit | query | Total rows to return. Default 100, max 500. |
| cursor | query | From next_page_url. |
Example
curl "https://dirigon.com/v1/filers/1287750/positions?limit=5" \ -H "X-API-KEY: YOUR_KEY"
import requests
headers = {"X-API-KEY": "YOUR_KEY"}
url = "https://dirigon.com/v1/filers/1287750/positions?limit=25"
r = requests.get(url, headers=headers)
for row in r.json()["positions"]:
print(row["company_raw"], row["lien"], row["mark"], row["fv_m"])
Response
{
"positions": [
{
"company_raw": "ACP Avenu Midco LLC",
"investment_raw": "First lien senior secured revolving loan",
"lien": "first_lien",
"mark": 1.0,
"fv_m": 0.9,
"cost_m": 0.9,
"non_accrual": false,
"filing_url": "https://www.sec.gov/Archives/edgar/data/1287750/000162828026050307/arcc-20260630.htm"
}
]
}
Fields
| Field | Meaning |
|---|---|
| company_raw | Company as printed on the SOI. |
| investment_raw | Investment line as printed. |
| lien | first_lien, second_lien, and so on. |
| cost_m | Amortized cost, $ millions. |
| fv_m | Fair value, $ millions. |
| mark | Fair value / cost. |
| non_accrual | True if the fund footnoted it as non-accrual. |
| filing_url | The filing this line came from. |
Non-accrual
What the fund footnoted. Not inferred from a low mark.
GET/v1/non-accrual
Example
curl "https://dirigon.com/v1/non-accrual?limit=10" \ -H "X-API-KEY: YOUR_KEY"
import requests
headers = {"X-API-KEY": "YOUR_KEY"}
r = requests.get("https://dirigon.com/v1/non-accrual?limit=10", headers=headers)
for row in r.json()["non_accrual"]:
print(row["filer"], row["company"], row["mark"])
Response
{
"non_accrual": [
{
"filer": "Blackstone Private Credit Fund",
"company": "Medallia, Inc.",
"cost_m": 1084.03,
"fv_m": 553.6,
"mark": 0.5107,
"period": "2026-06-30"
}
]
}
Coverage
How many funds and credits are in the latest snapshot.
GET/v1/coverage
No query parameters. Call this when you want the current universe size.
curl "https://dirigon.com/v1/coverage" \ -H "X-API-KEY: YOUR_KEY"
import requests
headers = {"X-API-KEY": "YOUR_KEY"}
print(requests.get("https://dirigon.com/v1/coverage", headers=headers).json())
MCP
Same data, for agents. Do not paste an API key into a chat.
Open Connect. Claude, ChatGPT, and Cursor use OAuth. Curl uses header X-API-KEY from an env var. An invite terminal seat cannot Allow.
Claude app
- Customize → Connectors → Add custom connector.
- Name:
Dirigon. URL:https://dirigon.com/v1/mcp - Leave OAuth Client ID and Secret empty. Click Add.
- Sign in with the mailbox on the API plan and click Allow.
- New chat → + → enable Dirigon.
Do not paste an API key into that dialog or into a chat.
Claude Code
claude mcp add --transport http dirigon https://dirigon.com/v1/mcp
Tools
lookup- First tool. q= a borrower or ticker. One short card.full=trueis the deep card.get_issuer- Peeled credit. One hit or exact name opens the card. q= is a whole token. preset=disagreement is the widest-marks screen.get_disagreement- Widest same-loan screen. Pass issuer_id to get_issuer for the card.get_filer- Fund card: book, top credits, software membership vs printed mixget_positions- Line lookup (cik required; q= a borrower; no q is a 15-line sample)get_non_accrual- Non-accrual lines (optional cik or q)get_coverage- Snapshot size, plan, and start_hereget_sector- Sectors page (software, healthcare, industrials)get_software- Same as get_sector for softwareget_clo_tranche- CLO note holders across CEFs, BDCs, and interval funds
Example
curl https://dirigon.com/v1/mcp \
-H "X-API-KEY: YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
Example prompts
- Who holds Coupa, and at what mark?
- Open BCRED. What is the book mark vs as-filed?
- Four largest funds, equal-weight: book mark, PIK, non-accrual.
- Which fund has the most software?
- Show the widest holder-mark disagreements in software.
- Show Ares Capital's Kellermeyer line.
- Which funds flagged Medallia as non-accrual?
OpenAPI
Generate a client, import into Postman, or let an agent discover the endpoints.
https://dirigon.com/v1/openapi.yaml
Full field list: /v1/fields.json. Agent index: /llms.txt. Markdown pages: /docs/introduction.md.