Execution Log: adopt an unconfigured repository before the session is spawned

Append-only log for issue-201. Ticket: #201.

Phase transitions

Phase	Entered	Reviewed/approved by	Notes
phase-selection	2026-08-10	@MadaraUchiha-314 (out of band)	Raised by the owner while reviewing #195 and worked in the same cloud session, so no checklist was posted. Scaled to the change per config.autonomy: a bugfix.md carrying its own design and test matrix in place of the four-file chain (decision-045 allows either name at the gate).
requirements-definition	2026-08-10		bugfix.md locked — two requirements: the config exists before the harness starts, and the gates do not move with it.
implementation	2026-08-10		GraphLink.adopt, two dispatcher call sites, the ordering test.
verification	2026-08-10		Every row of the matrix executed; evidence recorded.
needs-review	2026-08-10		Self-review; awaiting the human gate on the PR.

Pull requests

PR	Scope / tasks	Status
MadaraUchiha-314/the-loop — claude/github-issue-193-x96m3i	the whole work item	open

Progress entries

2026-08-10 — the fix

• Phase: requirements-definition → implementation → verification
• Did: Split issue-193's _adopt into GraphLink.adopt (public, pre-spawn, running the coupling's gates itself because cwd is not yet proved to be the work item's repository at that point) and _write_default (the shared writer, carrying the contribution carve-out), with _adopt left on the driving actions as an idempotent safety net. Called from Dispatcher._spawn_session between _prepare_workspace and the context read, and from the respawn path beside _prepare_environment — the two places a harness process is about to start in a checkout.
• Checkpoint/tests: the ordering test asserts from inside FakeTmux.spawn, and was verified red by disabling the pre-spawn call before being kept — an ordering test that passes with the fix reverted proves nothing. make test 1782 passed, 1 skipped.
• Next: the reviewer briefing on the PR.

Verification results

Recorded in bugfix.md § Testing, against the matrix rows that planned them. This work item has no separate testing-plan.md: the matrix is small enough to live with the requirements it proves.

Design critic review

Not selected. design-critic-review is opt-in (issue-188) and this work item did not tick it.

Review cycles

Cycle	Type (self/critic/security)	Reviewer	Outcome	Link
1	self	the-loop (this session)	new findings — the first cut left _adopt's three-part docstring describing a placement that had moved; split into adopt (ordering + gates) and _write_default (the carve-out) so each says what it actually does	graphlink.py
2	self	the-loop (this session)	new findings — _SeqLink, the dispatcher's GraphLink double, lacked the new method and failed three unrelated tests. Adding it to the double (rather than making the dispatcher tolerant of a missing method) keeps a wiring bug loud, and let the existing sequence assertion cover adoption too	test_graph_drive_integration.py
3	self	the-loop (this session)	zero (converged)	—
4	critic	—	unavailable — reviews.critics is empty in this project's config	.the-loop/harness-config.yaml
5	security	the-loop checklist	no findings — see the gate below	bugfix.md

Security review (gate)

• Mechanism: the-loop checklist (security.review.mechanism: auto; the built-in skill ran for issue-193 over the same code, and this work item moves a call rather than changing what it does).
• Outcome: pass, no findings. The write target, the owner/repo allow-list, the symlink containment and the never-overwrite rule are untouched. The one risk the move introduced was writing into a cwd not yet proved to be the work item's repository — which is why the ownership proof moved with the write instead of being left behind in _guarded. Recorded as bugfix.md § Security considerations with its two abuse cases.
• Human sign-off: n/a — risk tier 3, below security.review.humanSignOffMinTier: 4.

Final validation evidence

Requirement	Proved by
R1 — the config exists before the harness starts	test_the_repository_is_adopted_before_the_harness_is_started, asserted from inside the spawn call and verified red without the fix; test_a_spawn_reads_context_before_render_and_enters_after pins the full adopt → context → deliver → spawn order
R2 — the gates do not move with it	issue-193's suite unchanged and green (47 passed), including the foreign-checkout, contribution and never-overwrite scenarios

Capability docs

Capability doc	What changed	History row
webhook-triggers.md	The adoption bullet now states when it happens — before the prompt is rendered and before anything spawns — and names the respawn pre-flight	issue-201 row added at the top of § History

Documentation

Document	What changed
docs/config/harness-config.md	The When a repository has no config table now says the daemon adopts before the session starts, so a reader knows the guarantee, not just the behaviour
skills/the-loop/reference/automation.md	Same clarification in the bullet an agent working under the-loop reads
decision-073	Unchanged: the decision it records — one packaged default, one writer, the contribution carve-out — is exactly what this work item preserves. The ordering was an implementation defect, not a change of mind, so it is recorded here rather than as a new decision