Current - 2026/07
Endpoint
POST
/v1/stable/transcript/search

Finds recorded conversations matching a filter. Unlike the list calls elsewhere in this API, the body takes take and skip, so results arrive a page at a time.

Request
Response

Authentication

Authorization
string
header
required

Voiceflow bearer token

Query parameters

projectID
string
required

The ID of the project to operate on.

Request body

take
number
default:20

The maximum number of results to return, used for pagination.

Minimum: 1 · Maximum: 100
skip
integer
default:0

The number of results to skip, used for pagination.

Minimum: 0 · Maximum: 9007199254740991
endDate
string

When provided, only transcripts created at or before this ISO 8601 timestamp are returned.

version
enum<string>

When provided, only transcripts from the draft or published version of the environment are returned. published covers the current release and every past release. Applies within a single environment; the main environment is used when environmentAlias is omitted.

Available options: draft, published
filters
object
sessionID
string

When provided, only transcripts from this conversation session are returned.

startDate
string

When provided, only transcripts created at or after this ISO 8601 timestamp are returned.

environmentAlias
string

When provided, only transcripts from the environment with this alias are returned.

releaseName
string

When provided, only transcripts from the release with this exact name (e.g. V1.02) are returned, as listed in the releases of the environment. When several releases share the name, transcripts from all of them are returned. Applies within a single environment; the main environment is used when environmentAlias is omitted. Cannot be combined with version: "draft".

Minimum length: 1

Response

200 - application/json

Returns the transcripts matching the search, each with its session and user IDs, timestamps, recording URL, and any properties and evaluation results attached to it. The conversation logs are not included.

transcripts
object[]
required
StableTranscript — see its attributes above
Endpoint
POST
/v1/stable/transcript/grep

Find a pattern in the dialogue of many transcripts.

Request
Response

Authentication

Authorization
string
header
required

Voiceflow bearer token

Query parameters

projectID
string
required

The ID of the project to operate on.

Request body

pattern
string
required

Regex (case-insensitive) or plain text to find in the dialogue. A pattern may hold one open quantifier (*, +, ?, {n,} or {n,m}), no backreference and no alternation inside a group. Use exact counts such as \d{3}, or put the alternation at the top level: refund|return. Any other pattern runs as plain text, and the result says so.

Minimum length: 1
transcriptIDs
string[]

Transcripts to scan. Leave it out to scan the newest transcripts in the date range.

Minimum items: 1 · Maximum items: 60
startDate
object

Lower bound when no transcriptIDs are given: a day or a timestamp.

endDate
object

Upper bound when no transcriptIDs are given: a day or a timestamp.

environmentAlias
string

Only transcripts from this environment, when no transcriptIDs are given.

limit
integer
default:25

How many transcripts to scan at most, 1 to 60.

Minimum: 1 · Maximum: 60

Response

200 - application/json
pattern
string
required
scanned
integer
required

Transcripts scanned.

Minimum: -9007199254740991 · Maximum: 9007199254740991
matched
integer
required

Transcripts with at least one match.

Minimum: -9007199254740991 · Maximum: 9007199254740991
capped
boolean

More transcripts were in the range, or more unique transcriptIDs were given, than were scanned; note says what to do.

errored
integer

Transcripts whose log could not be read.

Minimum: -9007199254740991 · Maximum: 9007199254740991
skipped
integer

Given transcriptIDs that do not belong to this project.

Minimum: -9007199254740991 · Maximum: 9007199254740991
note
string
literalText
enum<boolean>

The pattern ran as plain text; note says why.

Available options: true
hits
object[]
required
StableTranscriptGrepHit — see its attributes above
Current - 2026/07
Endpoint
GET
/v1/stable/transcript/{transcriptID}

Returns one recorded conversation, including the turns exchanged and the traces behind them. Set filterConversation to keep only text, speak, and live-agent-handoff traces, plus any types listed in customTraceTypes.

Request
Response

Authentication

Authorization
string
header
required

Voiceflow bearer token

Path parameters

transcriptID
string
required

The ID of the transcript to operate on.

Query parameters

projectID
string
required

The ID of the project to operate on.

filterConversation
boolean

When true, only text, speak, and live-agent-handoff traces are returned. Omit it or set false to return the full log.

customTraceTypes
object

Extra trace types to keep when filterConversation is true. Ignored when the filter is off. Accepts one type or an array.

Response

200 - application/json

Returns a single transcript along with its logs, which the search operation leaves out, plus its properties, evaluation results, and recording URL.

transcript
object
required
Endpoint
GET
/v1/stable/transcript/{transcriptID}/files

Returns the files attached to one transcript that Voiceflow stores, each with a presigned download URL; files referenced only by an external URL are left out. The call is unpaginated: it takes no paging parameters, so one request yields the whole set.

Request
Response

Authentication

Authorization
string
header
required

Voiceflow bearer token

Path parameters

transcriptID
string
required

The ID of the transcript to operate on.

Query parameters

projectID
string
required

The ID of the project to operate on.

Response

200 - application/json

Returns the transcript's stored files, each with a stable ID, MIME type, original filename and size when recorded, a presigned GET URL, and the timestamp after which that URL stops working.

files
object[]
required

The files attached to the transcript that are stored on Voiceflow.

StableTranscriptFile — see its attributes above
Endpoint
POST
/v1/stable/transcript/{transcriptID}/logs

Searches one recorded conversation's log by entry kind, level, debug kind, turn or text, and returns the matches a page at a time. Use it to find errors or tool calls without reading the whole transcript.

Request
Response

Authentication

Authorization
string
header
required

Voiceflow bearer token

Path parameters

transcriptID
string
required

The ID of the transcript to operate on.

Query parameters

projectID
string
required

The ID of the project to operate on.

Request body

kinds
object[]

Keep only entries of these kinds. Omit to keep every kind.

Minimum items: 1 · Maximum items: 5
StableLogEntryKind — see its attributes above
levels
object[]

Keep only entries at these levels, for example ["fatal", "error", "warn"] for what went wrong.

Minimum items: 1 · Maximum items: 5
StableLogLevel — see its attributes above
debugKinds
object[]

Keep only debug entries of these kinds. Setting it leaves out every entry that is not a debug log.

Minimum items: 1 · Maximum items: 8
StableLogDebugKind — see its attributes above
turns
integer[]

Keep only entries of these turns. Turn 0 holds anything logged before the first input.

Minimum items: 1 · Maximum items: 100
search
string

Keep only entries whose message or recorded data contains this text, ignoring case. It is plain text, not a regular expression. Separate alternatives with |, as in "timeout|denied", to keep entries that contain any of them.

Minimum length: 1 · Maximum length: 256
detail
object
default:"summary"
limit
integer
default:50

The most entries to return in one page.

Minimum: 1 · Maximum: 200
offset
integer
default:0

How many matching entries to skip, for paging.

Minimum: 0 · Maximum: 9007199254740991
order
object
default:"asc"

Response

200 - application/json

Returns the matching entries for this page, with the number of matches, the whole log's entry counts and token use per model, and the offset to send for the next page.

total
integer
required

Entries in the whole log.

Minimum: -9007199254740991 · Maximum: 9007199254740991
turnCount
integer
required

The last turn the log reaches.

Minimum: -9007199254740991 · Maximum: 9007199254740991
counts
object
required

How many entries of each kind, level and debug kind the whole log holds, before any filter.

StableLogCounts — see its attributes above
tokens
object[]
required

Token use per model over the whole log, largest first.

StableLogTokenUsage — see its attributes above
matches
integer
required

Entries that pass every filter.

Minimum: -9007199254740991 · Maximum: 9007199254740991
entries
object[]
required

This page of the matching entries.

StableLogEntry — see its attributes above
hasMore
boolean
required
nextOffset
integer
required

The offset of the next page, or null when this page reaches the last match.

Minimum: -9007199254740991 · Maximum: 9007199254740991
truncated
boolean
required

True when the page stopped before limit to stay under about 30,000 tokens. Continue from nextOffset.

Schemas

Use the up and down arrow keys to select a result, Enter to open it, and Escape to close the search.