The GitHub adapter connects Sortie to GitHub Issues via the GitHub REST API (version 2026-03-10). It fetches candidate issues from the issues list endpoint (or the search endpoint when query_filter is configured), derives Sortie states from issue labels, normalizes responses to the domain issue model, paginates using Link header navigation, and maps HTTP errors to Sortie's normalized error categories. Registered under kind "github".

GitHub Enterprise Server is supported. Set endpoint to your GHES base URL. The sub-issue (parent) and dependency (blocked_by) endpoints are available on all GitHub plans; the adapter degrades gracefully to nil and [] respectively when the endpoints return 404.

See also: WORKFLOW.md configuration for the full tracker schema, error reference for all tracker error kinds, environment variables for $VAR expansion behavior.

Configuration¶

The adapter reads its configuration from the tracker section of the WORKFLOW.md front matter. Two fields are required; the rest have defaults.

endpoint¶

The GitHub API base URL. The default value is https://api.github.com. For GitHub Enterprise Server, set this to your instance's API root (for example, https://github.mycompany.com). Trailing slashes are stripped.

Accepts $VAR indirection when the entire value is a variable reference.

api_key¶

A GitHub personal access token (classic or fine-grained). This field is not in email:token format — the value is the token string alone.

Minimum required scopes for classic tokens: repo (reads issues, posts comments, manages labels).

Minimum required permissions for fine-grained tokens: Issues (read and write), Metadata (read).

Accepts $VAR indirection anywhere in the string via full os.ExpandEnv expansion.

api_key: $SORTIE_GITHUB_TOKEN
api_key: $GITHUB_TOKEN

project¶

Repository in owner/repo format — for example, myorg/myrepo. The adapter splits on the / to extract the owner and repository name. A value with zero or more than one /, or with empty parts, produces a tracker_payload_error at construction time.

project: myorg/myrepo
project: $SORTIE_GITHUB_PROJECT

active_states¶

Label names that map to active Sortie states. Issues with one of these labels are eligible for dispatch. Values are compared case-insensitively and stored lowercased at construction time.

When omitted, defaults to ["backlog", "in-progress", "review"]. These label names must exist in the repository — GitHub has no built-in equivalents.

terminal_states¶

Label names that map to terminal Sortie states. Issues with one of these labels trigger workspace cleanup. Stored lowercased.

When omitted, defaults to ["done", "wontfix"].

query_filter¶

A raw GitHub search qualifier string. When this field is non-empty, FetchCandidateIssues switches from the issues list endpoint to the search endpoint and appends this value to the base query repo:{owner}/{repo} type:issue state:open.

query_filter: "label:agent-ready"
query_filter: "label:agent-ready milestone:v2"

Do not include repo: or type:issue in the value — they are added automatically.

Pre-creating labels¶

The adapter does not create issue labels automatically. All label names referenced in active_states, terminal_states, handoff_state, and in_progress_state must exist in the repository before Sortie starts. A TransitionIssue call that references a non-existent label produces a tracker_payload_error (HTTP 422 from GitHub).

Validate-time checks¶

When tracker.kind is github, the sortie validate pipeline runs GitHub-specific config checks in addition to the generic preflight validation. These checks run without constructing an adapter instance or making network calls.

Errors¶

Empty tracker.project is caught by the generic preflight check (tracker.project is required) before adapter validation runs.

Warnings¶

The api_key warnings are supplementary hints. The generic preflight check already reports an error when tracker.api_key is empty — the adapter-specific warnings provide actionable remediation guidance alongside that error.

Authentication¶

Every request sets a Bearer authorization header:

Authorization: Bearer <token>

Additional fixed headers on all requests:

The HTTP client has a 30-second per-request timeout. Context cancellation is propagated — a cancelled context causes the in-flight request to return immediately with context.Canceled.

State derivation¶

GitHub issues have two native states: open and closed. Sortie states are derived from issue labels using a four-priority algorithm.

Priority order¶

1. Active states, config order. Issue labels are scanned against active_states in configuration order. The first match is returned.
2. Terminal states, config order. If no active state matched, labels are scanned against terminal_states in configuration order. The first match is returned.
3. Native-state fallback. If no label matched either list:
4. open issue → active_states[0] (first configured active state, e.g., "backlog").
5. closed issue → terminal_states[0] (first configured terminal state, e.g., "done").
6. Native state passthrough. When both active_states and terminal_states are empty (not recommended), returns "open" or "closed" directly.

Multi-label conflicts¶

When an issue carries multiple state labels, the first configured active state wins (priority 1). Configuration order is deterministic; label display order on the issue is irrelevant.

Unlabeled issues¶

An open issue with no state label resolves to active_states[0]. A closed issue resolves to terminal_states[0]. This prevents unlabeled issues from appearing as an unknown state in the orchestrator.

Case handling¶

All comparisons are case-insensitive. A label named "In-Progress" matches the configured value "in-progress". All stored and compared values are lowercased at construction time.

API operations¶

The adapter implements all eight methods of the TrackerAdapter interface.

FetchCandidateIssues¶

Returns issues in configured active states.

When query_filter is empty (default — issues endpoint):

• Endpoint: GET /repos/{owner}/{repo}/issues
• Parameters: state=open, sort=created, direction=asc, per_page=50
• All open issues are fetched and filtered client-side by state label. Pull requests are filtered out via the pull_request field marker.

When query_filter is set (search endpoint):

• Endpoint: GET /search/issues
• Query (q): repo:{owner}/{repo} type:issue state:open {query_filter}
• Parameters: sort=created, order=asc, per_page=50
• incomplete_results: true in the response body produces a WARN log but does not abort the request.

Pagination: Link header-based (rel="next"). Page size: 50. Maximum 200 pages (10,000 issues). When the page limit is reached, a WARN is logged and accumulated results are returned without error.

Comments: Set to nil on returned issues.

Pull request filtering: Applied to all responses. The pull_request field on a list entry is non-nil for pull requests; those entries are skipped.

FetchIssueByID¶

Returns a single fully-populated issue. The issueID parameter is the issue number as a string, equal to the Identifier field.

Four requests:

1. GET /repos/{owner}/{repo}/issues/{issueID} — issue body and labels.
2. GET /repos/{owner}/{repo}/issues/{issueID}/dependencies/blocked_by — blocker list. Returns [] on 404.
3. GET /repos/{owner}/{repo}/issues/{issueID}/parent — parent issue. Returns nil on 404.
4. GET /repos/{owner}/{repo}/issues/{issueID}/comments — comments, Link-header paginated.

Returns tracker_not_found when the issue does not exist (HTTP 404) or when the resolved entity is a pull request.

FetchIssuesByStates¶

Returns issues in specified Sortie states. Used for startup terminal cleanup.

Active states: Issues endpoint with state=open, client-side label filtering.

Terminal states: Search endpoint, one query per terminal-state label: repo:{owner}/{repo} type:issue state:closed label:{terminal_label}. Server-side filtering avoids scanning all closed issues in the repository.

Returns an empty slice when states is empty.

FetchIssueStatesByIDs¶

Returns the current state for each requested issue ID. Since ID == Identifier == issue number for this adapter, the IDs are used directly as issue numbers in individual API calls.

Endpoint (per issue): GET /repos/{owner}/{repo}/issues/{number}

Batching: None. Sequential individual requests. 404 responses are omitted from the result map without error.

FetchIssueStatesByIdentifiers¶

Structurally identical to FetchIssueStatesByIDs. Since ID == Identifier for this adapter, both methods share the same internal implementation.

FetchIssueComments¶

Returns all comments for an issue.

Endpoint: GET /repos/{owner}/{repo}/issues/{issueID}/comments

Pagination: Link header-based. Page size: 50. Maximum 200 pages.

Returns tracker_not_found when the issue does not exist.

TransitionIssue¶

Applies a state transition by manipulating issue labels and the open/closed native state.

Steps:

1. GET /repos/{owner}/{repo}/issues/{issueID} — read current labels and native state.
2. DELETE /repos/{owner}/{repo}/issues/{issueID}/labels/{old_label} — remove the current state label, if present and different from the target. Label names are URL path-escaped. A 404 here is treated as a no-op (label already absent).
3. POST /repos/{owner}/{repo}/issues/{issueID}/labels — add the target state label.
4. If the target is a terminal state and the issue is open: PATCH /repos/{owner}/{repo}/issues/{issueID} with {"state": "closed", "state_reason": "completed"}.
5. If the target is an active state and the issue is closed: PATCH /repos/{owner}/{repo}/issues/{issueID} with {"state": "open"}.

Atomicity: The steps are not atomic. A failure at any step causes the adapter to return an error; the orchestrator retries on the next tick. Label operations are idempotent — retries converge to the correct state without creating duplicates.

Label case: Target labels are sent as configured (lowercased). GitHub label matching is case-insensitive.

CommentIssue¶

Posts a plain-text comment on an issue.

Endpoint: POST /repos/{owner}/{repo}/issues/{issueID}/comments

Request body: {"body": "<text>"}. No ADF conversion — GitHub natively accepts Markdown.

Returns nil on success (HTTP 201).

Field mapping¶

ID and Identifier¶

Both ID and Identifier map to the GitHub issue number. The global integer id field returned by the API is not used as the adapter's ID — it cannot be used to look up issues via the REST API. As a result, FetchIssueStatesByIDs and FetchIssueStatesByIdentifiers are structurally equivalent for this adapter.

Comment normalization¶

Error mapping¶

403 disambiguation¶

GitHub uses HTTP 403 for both permission errors and secondary rate limits. The adapter applies a three-step check in order:

1. If the x-ratelimit-remaining header equals "0" → tracker_api_error (primary rate limit).
2. If the response body (up to 512 bytes) contains "rate limit" (case-insensitive) → tracker_api_error (secondary rate limit).
3. Otherwise → tracker_auth_error (insufficient permissions).

The Retry-After header value from 429 responses is included in the error message for diagnostics.

For the full error taxonomy and operator guidance, see the error reference.

Pagination¶

All list endpoints use Link header-based pagination.

The adapter follows rel="next" links directly — it does not construct URLs manually. A maximum of 200 pages are fetched per operation. When the limit is reached, accumulated results are returned with a WARN log.

Rate limits¶

GitHub enforces two independent rate limit buckets.

At the default 30-second poll interval with max_concurrent_agents: 10, typical usage is well within the primary rate limit. The search budget applies only when query_filter is configured or during the one-time startup terminal-state cleanup.

Rate limit violations return HTTP 429 or HTTP 403. Both are mapped to tracker_api_error. The orchestrator logs the error and waits for the next poll interval.