aicodingsession

package
v1.114.0 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Oct 2, 2026 License: Apache-2.0 Imports: 6 Imported by: 0

Documentation

Index

Constants

View Source 
const (
	// EvidenceID is the identifier for the AI coding session material type
	EvidenceID = "CHAINLOOP_AI_CODING_SESSION"
	// EvidenceSchemaURL is the URL to the JSON schema for AI coding session
	EvidenceSchemaURL = "https://schemas.chainloop.dev/aicodingsession/0.1/ai-coding-session.schema.json"
)
View Source 
const (
	// ModeCoding is a session traced through the agent and git hooks installed
	// by `chainloop trace init`. It is what an absent mode means, so sessions
	// recorded before the field existed keep their meaning.
	ModeCoding = "coding"
	// ModeGeneric is a single-shot session wrapped by `chainloop trace run`.
	ModeGeneric = "generic"
)

How `chainloop trace` was driven for a session. Recorded so that consumers can tell an ongoing coding session apart from a one-shot wrapped command instead of presenting them as the same thing. More modes are expected, so the field is a plain string rather than a closed enumeration.

View Source 
const (
	// SpecKindTicket is an issue tracker item: a Linear or Jira ticket.
	SpecKindTicket = "ticket"
	// SpecKindDocument is a written specification: a design doc, a page in a
	// vault, an RFC.
	SpecKindDocument = "document"
	// SpecKindImage is a spec given as a picture — a mockup or a screenshot —
	// of which the stored content is whatever the agent transcribed.
	SpecKindImage = "image"
	// SpecKindText is a spec stated in the session itself rather than resolved
	// from somewhere, and the fallback for anything unrecognised.
	SpecKindText = "text"
)

What a captured spec entry was resolved from, as a closed vocabulary that consumers switch on. Unlike Mode, this value originates in a model's output rather than in a Chainloop release, so a value outside the set is a mistake to normalise away rather than a future kind to pass through. The schema nonetheless leaves the field an open string, so a kind added by a later CLI is not rejected by a control plane that predates it.

Variables

This section is empty.

Functions

func Redact added in v1.107.0 

func Redact(ctx context.Context, evidence []byte) ([]byte, *redaction.Report, error)

Redact removes detected secrets from an AI coding session evidence document.

It returns the sanitised bytes and a summary of what was replaced. A document with no detected secrets is returned verbatim, so that the common case keeps its original digest. The result is guaranteed to still validate against the AI coding session schema; if it does not, redaction fails rather than uploading either an invalid document or an unredacted one.

func RedactSpecText added in v1.114.0 

func RedactSpecText(ctx context.Context, text string) (string, *redaction.Report, error)

RedactSpecText removes detected secrets from the text of one captured spec.

A spec is stored as its own material rather than inside the session document, so it does not pass through Redact. It gets the same scanner and the same placeholders here, so a credential pasted into a ticket body is removed exactly as it would be from the transcript. Text with no detected secrets is returned unchanged.

func ResolveMode added in v1.109.5 

func ResolveMode(mode string) string

ResolveMode maps an unset mode onto the one an absent mode means, so that the default is written down once instead of at each producer and consumer.

func ResolveSpecKind added in v1.114.0 

func ResolveSpecKind(kind string) string

ResolveSpecKind maps a captured kind onto the vocabulary above, so the normalisation is written down once instead of at each producer and consumer.

Types

type Agent 

type Agent struct {
	Name    string `json:"name"`
	Version string `json:"version,omitempty"`
}

Agent identifies the AI agent provider.

type CodeChanges 

type CodeChanges struct {
	FilesModified     int          `json:"files_modified,omitempty"`
	FilesCreated      int          `json:"files_created,omitempty"`
	FilesDeleted      int          `json:"files_deleted,omitempty"`
	LinesAdded        int          `json:"lines_added,omitempty"`
	LinesRemoved      int          `json:"lines_removed,omitempty"`
	AILinesAdded      int          `json:"ai_lines_added,omitempty"`
	AILinesRemoved    int          `json:"ai_lines_removed,omitempty"`
	HumanLinesAdded   int          `json:"human_lines_added,omitempty"`
	HumanLinesRemoved int          `json:"human_lines_removed,omitempty"`
	Files             []FileChange `json:"files,omitempty"`
}

CodeChanges summarizes code modifications made during the session.

type Conversation 

type Conversation struct {
	TotalMessages     int `json:"total_messages,omitempty"`
	UserMessages      int `json:"user_messages,omitempty"`
	AssistantMessages int `json:"assistant_messages,omitempty"`
}

Conversation holds message count statistics.

type Data 

type Data struct {
	SchemaVersion string  `json:"schema_version"`
	Agent         Agent   `json:"agent"`
	Session       Session `json:"session"`
	// Spec is what the session was asked to build, one entry per source the
	// agent resolved. Empty for a session that captured none.
	Spec         []SpecEntry                  `json:"spec,omitempty"`
	GitContext   *GitContext                  `json:"git_context,omitempty"`
	CodeChanges  *CodeChanges                 `json:"code_changes,omitempty"`
	Model        *Model                       `json:"model,omitempty"`
	Usage        *Usage                       `json:"usage,omitempty"`
	ToolsUsed    *ToolsUsed                   `json:"tools_used,omitempty"`
	Conversation *Conversation                `json:"conversation,omitempty"`
	Subagents    []Subagent                   `json:"subagents,omitempty"`
	RawSession   map[string][]json.RawMessage `json:"raw_session,omitempty"`
	Warnings     []string                     `json:"warnings,omitempty"`
}

Data is the AI coding session payload.

type Evidence 

type Evidence struct {
	ID     string `json:"chainloop.material.evidence.id"`
	Schema string `json:"schema"`
	Data   Data   `json:"data"`
}

Evidence represents the complete evidence structure for AI coding session.

func NewEvidence 

func NewEvidence(data Data) *Evidence

NewEvidence creates a new Evidence instance.

type FileChange 

type FileChange struct {
	Path         string      `json:"path"`
	Status       string      `json:"status"`
	LinesAdded   int         `json:"lines_added,omitempty"`
	LinesRemoved int         `json:"lines_removed,omitempty"`
	Attribution  string      `json:"attribution,omitempty"`
	LineRanges   []LineRange `json:"line_ranges,omitempty"`
	SessionIDs   []string    `json:"session_ids,omitempty"`
}

FileChange represents a single file modification in the session.

type GitContext 

type GitContext struct {
	Repository  string   `json:"repository,omitempty"`
	Branch      string   `json:"branch,omitempty"`
	WorkDir     string   `json:"work_dir,omitempty"`
	CommitStart string   `json:"commit_start,omitempty"`
	CommitEnd   string   `json:"commit_end,omitempty"`
	Commits     []string `json:"commits,omitempty"`
	CommitCount int      `json:"commit_count,omitempty"`
}

GitContext holds repository and commit information at capture time.

type LineRange added in v1.89.11 

type LineRange struct {
	Start int `json:"start"`
	End   int `json:"end"`
}

LineRange represents a contiguous range of lines.

type Model 

type Model struct {
	Primary    string   `json:"primary,omitempty"`
	Provider   string   `json:"provider,omitempty"`
	ModelsUsed []string `json:"models_used,omitempty"`
}

Model holds information about the AI models used in the session.

type Session 

type Session struct {
	ID   string `json:"id"`
	Slug string `json:"slug,omitempty"`
	// Mode is one of the Mode* constants. Empty means ModeCoding.
	Mode            string `json:"mode,omitempty"`
	StartedAt       string `json:"started_at"`
	EndedAt         string `json:"ended_at,omitempty"`
	DurationSeconds int    `json:"duration_seconds"`
}

Session holds timing and identity information for the coding session.

type SpecEntry added in v1.114.0 

type SpecEntry struct {
	// Kind is one of the SpecKind* constants.
	Kind string `json:"kind"`
	// URI is where the text came from. Empty when the task was stated in the
	// session itself and there is no external source to point at.
	URI string `json:"uri,omitempty"`
	// Digest identifies the EVIDENCE material holding the spec text, as
	// "sha256:<hex>".
	Digest string `json:"digest"`
	// CapturedAt is when the agent wrote this entry, RFC3339. It comes from the
	// file's modification time, so it is the last write rather than the first,
	// and it is trivially forgeable: not a trusted timestamp.
	CapturedAt string `json:"captured_at"`
}

SpecEntry is one source a coding session was built from: the ticket, document or prompt that set the task, resolved by the agent. It is what the work gets judged against, which no amount of diff can answer on its own.

The entry is a reference. The text itself is a separate EVIDENCE material in the same attestation, found by Digest, so that a source shared by many sessions is stored once and can be fetched and verified on its own.

type Subagent added in v1.89.11 

type Subagent struct {
	ID          string         `json:"id"`
	Type        string         `json:"type"`
	Description string         `json:"description"`
	Tokens      SubagentTokens `json:"tokens"`
}

Subagent describes a spawned subagent within the session.

type SubagentTokens added in v1.89.11 

type SubagentTokens struct {
	Input  int `json:"input"`
	Output int `json:"output"`
}

SubagentTokens holds token usage for a subagent.

type ToolSummary 

type ToolSummary struct {
	ToolName        string `json:"tool_name"`
	InvocationCount int    `json:"invocation_count"`
}

ToolSummary represents usage statistics for a single tool.

type ToolsUsed 

type ToolsUsed struct {
	Summary          []ToolSummary `json:"summary,omitempty"`
	TotalInvocations int           `json:"total_invocations,omitempty"`
}

ToolsUsed summarizes tool usage during the session.

type Usage 

type Usage struct {
	InputTokens              int     `json:"input_tokens,omitempty"`
	OutputTokens             int     `json:"output_tokens,omitempty"`
	TotalTokens              int     `json:"total_tokens,omitempty"`
	CacheReadInputTokens     int     `json:"cache_read_input_tokens,omitempty"`
	CacheCreationInputTokens int     `json:"cache_creation_input_tokens,omitempty"`
	EstimatedCostUSD         float64 `json:"estimated_cost_usd,omitempty"`
}

Usage holds token usage and cost information.

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.