activity-core/workplans/ACTIVITY-WP-0022-issuesink-no-default-forgejo.md

id	type	title	domain	repo	status	owner	topic_slug	created	updated	state_hub_workstream_id
ACTIVITY-WP-0022	workplan	IssueSink: stop default silent Forgejo issues for internal findings	infotech	activity-core	finished	codex	activity-core	2026-07-21	2026-07-21	d8e4a0c0-dc63-486b-b153-0292e63dba54

IssueSink: stop default silent Forgejo issues for internal findings

Origin

Filed from issue-core ISSUE-WP-0004-T05 after the work-record canon alignment (CUST-WP-0060, founder-reviewed architecture draft §4.2).

Incident backdrop: daily-todo-md-stale-review (and any definition whose rules call emit_tasks) treated IssueSink → issue-core → Forgejo as the default always-on path. That produced tracker issues nobody was meant to see as fleet coordination. Forgejo issue tracking is not the fleet coordination mechanism; work records originate as repo files (ADR-001).

issue-core has retargeted its own docs to connector, not landing zone (INTENT.md / SCOPE.md / ISSUE-WP-0004). This workplan is the activity-core-side product fix: the default sink must not silently create Forgejo issues for internal findings.

Related:

• issue-core: workplans/ISSUE-WP-0004-align-with-work-record-canon.md
• the-custodian: CUST-WP-0060 closure review item 6 (TODO-stale / IssueSink)
• activity-core: ACTIVITY-WP-0021 (prod reliability; T01 may restore or replace emit path — this WP owns the policy default)
• src/activity_core/issue_sink.py — ISSUE_SINK_TYPE defaults to "state-hub" (ACTIVITY-WP-0022)

Goal

1. Default is safe for internal findings: no silent Forgejo issue spam when ISSUE_SINK_TYPE is unset or when a definition is internal-only.
2. Intentional external issues remain possible (explicit config / sink type / per-definition opt-in).
3. Internal outcomes go to the appropriate fleet surface (work-record intake, State Hub progress, working-memory, etc.) — not issue-core by default.
4. Document the sink matrix for operators (rest / null / future intake).

Out of scope

• Implementing issue-core UUID ↔ external-id mapping (issue-core design note docs/uuid-external-id-mapping.md; stage-3 work-record arch).
• Full redesign of every ActivityDefinition — only sink policy + the definitions that currently misuse IssueSink for internal findings.
• ACTIVITY-WP-0021 bridge 503 reliability work (coordinate, don't duplicate).

Tasks

Task: Decide default sink policy and config surface

id: ACTIVITY-WP-0022-T01
status: done
priority: high
state_hub_task_id: "1a0d9bf6-c822-42c8-9b67-8a9be16cce7d"

Choose and document:

• Safe default for ISSUE_SINK_TYPE (e.g. null or a future non-Forgejo sink) vs keep rest only when explicitly enabled in prod overlays.
• Per-definition or per-rule opt-in for external projection.
• Interaction with NullSink, tests, and railiance01 deployment env.

Reference: issue-core SCOPE “Default sink for internal automation findings” and ISSUE-WP-0004 origin.

Task: Implement default that does not create Forgejo issues

id: ACTIVITY-WP-0022-T02
status: done
priority: high
state_hub_task_id: "78f994f6-68bd-4f18-88b0-3a25b46e7563"

Code + config changes so unmatched / internal-finding paths cannot spam Forgejo. Update tests (tests/test_issue_sink.py and callers of get_issue_sink()). Preserve a documented path for intentional rest use.

Task: Reroute or pause definitions that abused IssueSink for fleet work

id: ACTIVITY-WP-0022-T03
status: done
priority: medium
state_hub_task_id: "d45f861d-4c5e-4721-8e02-50d7deafaefd"

At minimum: daily-todo-md-stale-review and any other definition that only needs internal attention. Prefer work-record intake / State Hub progress over Forgejo. Coordinate with CUST-WP-0060 item 6 and WP-0021 if still paused.

Task: Operator docs and deployment env audit

id: ACTIVITY-WP-0022-T04
status: done
priority: medium
state_hub_task_id: "747a19ab-c6a5-4f8a-b294-a8dc0f82109f"

Document sink matrix; audit railiance01 / compose env for ISSUE_SINK_TYPE, ISSUE_CORE_URL, ISSUE_CORE_API_KEY. Confirm defaults match policy after T02.

Acceptance

• Unset / default config does not open Forgejo issues for internal findings
• Intentional external emit still works when explicitly configured
• TODO-stale (and peers) no longer depend on IssueSink→Forgejo for fleet visibility
• Tests and deploy docs match the new default
• Cross-link from issue-core ISSUE-WP-0004 remains accurate

See also

• src/activity_core/issue_sink.py
• issue-core INTENT.md (pivot 2026-07-20)
• the-custodian/canon/standards/work-record-types_v0.1.md
• the-custodian/research/WorkOrchestrationArchitectureDraft.md §4.2

Closeout 2026-07-21

Policy decision (T01): ISSUE_SINK_TYPE default = state-hub. rest is explicit opt-in only. Unknown/empty → state-hub.

Code (T02): DEFAULT_ISSUE_SINK_TYPE, factory fallback, tests.

Definitions (T03): daily-todo-md-stale-review already paused/disabled in prod; SBOM task rules gated off; internal paths use reports / state-hub.

Docs (T04): docs/issue-core-emission-boundary.md sink matrix, docs/task-emission-consumer-contract.md, .env.example, runbook.

Acceptance checkboxes:

• Default does not open Forgejo issues for internal findings
• Intentional rest still available
• TODO-stale not depending on IssueSink→Forgejo
• Tests and deploy docs match