triage(#135): dashboard tooltip → /bicameral-sync (scope-cut from auto-resolve)

[jinhongkuan]

Summary

Triage scope-cut for #135 (post-commit hook auto-resolve loop, P0). After
design enumeration of 7 options, every path to auto-resolution either (a)
requires an LLM in the deterministic core — violating the selection over
generation guardrail — or (b) requires a trivial-cases enumeration with
non-zero false-positive risk against the integrity contract.

Cut: accept the architectural limit. Post-commit hook stays sync-only.
Resolution path = dashboard tooltip on status === 'pending' rows → user
runs /bicameral-sync in their Claude Code session. The qualified judge of
any pending compliance check is the human-in-session, not a hook-side
automaton.

Changes

File	Δ	Notes
pilot/mcp/assets/dashboard.html	+5/-1	renderStateCell() ternary → if/else if; new pending branch with tooltip
pilot/mcp/skills/bicameral-dashboard/SKILL.md	+1/-0	One bullet under Notes documenting tooltip nudge contract
pilot/mcp/docs/META_LEDGER.md	+252/-3	Entries #24 (Audit PASS) + #25 (IMPL) + #26 (SEAL)
pilot/mcp/docs/SYSTEM_STATE.md	+72/-0	#135 triage substantiation addendum

Tooltip text: "Pending compliance — run /bicameral-sync in your Claude Code session to resolve."

Section 4 razor: renderStateCell 19 LOC (cap 40), nesting 1 (cap 3),
nested ternaries 0 (replaced ternary with if/else if — improves razor score).

Test plan

Mandatory manual verification — no automated UI test exists for
pilot/mcp/assets/dashboard.html; this is the agreed mitigation per Entry
#24 audit advisory:

• cd pilot/mcp && python -m server (or equivalent dev-server invocation); open dashboard URL.
• Ingest a decision, bind it to a code region (/bicameral-ingest + bicameral.bind), modify the bound file, git commit.
• In the dashboard, observe the row for that decision: status badge shows … pending; on hover, a tooltip appears with the text above.
• Run /bicameral-sync in a Claude Code session — tooltip's call-to-action resolves the pending state to reflected (or drifted).
• If hover yields no tooltip, change did not land — block merge.

Existing test suite: unaffected (no Python/server code touched).

Chain integrity

• Audit verdict: PASS, L1 — Entry fix: FC-2 follow-ups — symbol seeding, team-mode test unwrap, step-1 parser  #24 chain 1de1fac7…
• Implementation — Entry fix: Stage 1 symbol ranking by fuzzy score — MRR@3 0.39→0.58  #25 chain 51c8a45c…
• Substantiation seal — Entry fix: compound token extraction for grounding recall (9%→14%) #26 chain efd0304b…
• 26 entries on this branch, chain valid.

Post-merge external gh actions (queued)

Will fire after merge: gh issue close BicameralAI/bicameral-mcp#135 --comment ...

Closing as scope-cut. Auto-resolve direction abandoned after design
enumeration: no caller-LLM is reliably available in the post-commit hook
context, and MCP sampling is not surfaced in Claude Code's main chat
session.

Resolution path adopted: dashboard surfaces a tooltip on
`status === 'pending'` rows nudging the user to run `/bicameral-sync` in
their Claude Code session. The tooltip ships in this PR (#XXX). The
qualified judge of any pending compliance check is the
human-in-session, not a hook-side automaton.

README/docs deferral status: **N/A** — the original direction
(`--auto-resolve-trivial` flag, two-phase sync→resolve README section,
`docs/architecture/drift-resolution-loop.md`, CHANGELOG migration note)
never landed, so there are no docs to update or remove. Future state is
already accurate.

If the architectural premise changes (Claude Code exposes MCP sampling
reliably in the main chat, or a Hosted Repo / GitHub App tier ships), a
new issue should track that direction — not this one.

Will fire after merge: gh issue edit BicameralAI/bicameral#108 --body-file -

Two sentence-level spec corrections:

Flow 3 — append at end:

If a commit happens outside an active Claude Code session,
pending-compliance accumulates silently. The dashboard surfaces a
tooltip on status === 'pending' rows nudging the user to run
/bicameral-sync in a fresh session to walk the queue.

Flow 1 step 2 — replace:

bicameral.ingest → normalize payload, write ledger; returns
supersession_candidates and context_for_candidates

with:

bicameral.ingest → normalize payload, write ledger; returns
context_for_candidates. Supersession detection is performed
caller-side by bicameral-context-sentry (Probe B), which writes
signoff.state='collision_pending' for newly-ingested decisions that
overlap existing ones; subsequent bicameral.preflight calls surface
these via unresolved_collisions.

Out of scope (deferred to separate PRs)

• sim_issue_108_flows.py — simulation against the corrected Add deterministic escalation policy engine for semantic drift outcomes #108 spec, runs on dev after this triage merges.
• docs cleanup — remove stale pilot/mcp/skills/ references across CLAUDE.md, DEV_CYCLE.md, TODO.md #125 — stale pilot/mcp/skills/ reference cleanup. Finding surfaced during this triage: scope of docs cleanup — remove stale pilot/mcp/skills/ references across CLAUDE.md, DEV_CYCLE.md, TODO.md #125 should widen — 7 skills under pilot/mcp/.claude/skills/ are absent from canonical pilot/mcp/skills/ (per pilot/mcp/CLAUDE.md claim).
• #122, label-merged-to-dev workflow fails with 'Resource not accessible by integration' on some PRs #115 — CI infra bugs, separate focused PRs.

Risk

L1 — UI-only behavior change to existing infrastructure, plus issue/spec doc edits. No new code paths, no schema changes, no contract changes. Three audit preconditions (manual verification step, SKILL.md note, deferral-status close comment) all satisfied or composed.

Closes #135 (post-merge via queued gh issue close above).

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 5304a41f-dfdc-40e2-b890-82dcf7c49772

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch triage/135-dashboard-tooltip-scope-cut

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}