feat(agent): persist snapshot provenance counterfactuals and replay

[enjoyandlove]

Summary

Closes #270

• feat(agent): persist recommendation snapshot ids #281 — Added decision_snapshot_id column to agent_actions; every action produced by the orchestrator now carries a stable link to the context snapshot that drove it.
• feat(agent): add decision snapshot provenance fields #282 — Added provenance fields (actor_login, decision_pack_generated_at, confidence_level, freshness_at_decision, upstream_ruleset_id) to agent_context_snapshots so any decision can be inspected after the fact.
• test(agent): add public/private snapshot serialization fixtures #283 — Added test/fixtures/decision-snapshots/private.json and public-safe.json to pin the public/private serialization boundary; tests assert forbidden fields (payload, recommendationEvidence, alternativesConsidered, counterfactualReasons, reward language) never reach public output.
• feat(agent): add counterfactual reasons to decision snapshots #284 — Added alternatives_considered_json and counterfactual_reasons_json to agent_actions; orchestrator populates these from the decision pack — listing other ranked repos and the conditions that would flip the recommendation.
• feat(app): display decision snapshot replay details #285 — Added GET /v1/agent/snapshots/:snapshotId endpoint that returns a replayable { snapshotId, replayedAt, run, context, actions } payload, auth-gated to the original actor.

Scope

• This PR is focused and does not mix unrelated backend, UI, MCP, docs, dependency, and deploy changes.
• This follows CONTRIBUTING.md and does not reintroduce GitHub Pages, VitePress, site/, or CNAME.
• I linked an issue, or this is small enough that the summary explains why an issue is not needed.

Validation

• git diff --check
• npm run actionlint — skipped (network unavailable in dev environment: read ECONNRESET)
• npm run typecheck
• npm run test:coverage — 801 tests pass, branch coverage 97.01% (above threshold)
• npm run test:workers
• npm run build:mcp
• npm run test:mcp-pack
• npm run ui:openapi:check
• npm run ui:lint
• npm run ui:typecheck
• npm run ui:build
• npm audit --audit-level=moderate
• New or changed behavior has unit/integration tests for new branches, fallback paths, and sanitizer boundaries

If any required check was skipped, explain why:

• npm run actionlint — network reset error in dev environment; no workflow files were modified.
• npm run test:workers, build:mcp, test:mcp-pack, ui:* — no MCP package or UI files were changed; these gates are unchanged.

Safety

• No secrets, wallet details, hotkeys, coldkeys, user PATs, private keys, raw trust scores, private rankings, or private maintainer evidence are exposed.
• Public GitHub text stays sanitized, low-noise, and does not imply compensation guarantees or optimization tactics.
• Auth, cookie, CORS, GitHub App, Cloudflare, or session changes include negative-path tests. — N/A, no auth changes.
• API/OpenAPI/MCP behavior is updated and tested where needed. — new GET /v1/agent/snapshots/:snapshotId route tested with 404 and 200 paths.
• UI changes use live API data or real empty/error/loading states, not production mock/demo fallbacks. — N/A, no UI changes.
• Visible UI changes include screenshots or a short recording. — N/A, no UI changes.
• Public docs/changelogs are updated where needed. — N/A, internal API only.

Notes

• upstreamRulesetId column exists in schema and type but is not yet wired to a data source in the orchestrator. Populating it requires an additional DB query in the hot path; left as a follow-up. The field is nullable and safe to leave null until that wiring is added.
• confidence_level is nullable in the migration (ALTER TABLE ADD COLUMN in SQLite cannot be NOT NULL without a default); existing rows will have NULL, all new rows are always assigned a value by the orchestrator.
• alternativesConsidered and counterfactualReasons are optional (string[] | undefined) on AgentActionRecord for backward compatibility with test fixtures and inline action objects in existing tests.

[JSONbored]

@enjoyandlove this is moving in the right direction, but it is currently too broad for the split snapshot roadmap.

Required changes:

• Split the work against the smaller child issues instead of closing the whole snapshot epic in one PR.
• If /v1/agent/snapshots/:snapshotId remains in scope, update OpenAPI schemas/spec and generated UI artifacts.
• Keep the PR to one focused slice.

Validation expected:

• npm run test:ci
• npm run ui:openapi:check

[JSONbored]

@enjoyandlove this is mechanically blocked right now.

A few notes:

• The snapshot provenance/replay surface is large and needs green validation before deeper review.
• Current CI fails in typecheck at src/types.ts(712,1).
• Superagent has now passed, so validate is the remaining check blocker.

Required changes:

• Fix merge conflicts.
• Ensure CI is fully green.

Validation expected:

• npm run typecheck
• Focused decision snapshot tests
• Full npm run test:ci

[JSONbored]

@enjoyandlove thanks for the work here.

A few notes:

• This PR has become too stale and broad to keep carrying forward as the snapshot path evolves.
• The migration numbering is obsolete, and focused pieces of the work are already split into newer PRs.

Required changes:

• I would close this PR rather than keep rebasing it.
• If replay still needs a follow-up, open a fresh PR with only that remaining behavior and current migration numbering.

Validation expected:

• A replacement should include current migration order, focused replay tests, and the current PR-template evidence.