Custom detail command

fornax detail prints full evidence/finding detail for recent sessions — the drill-down behind the compact fornax status line.

Usage

fornax detail

🛡 ✕ CONTRADICTED
  claim:     All tests passed
  rationale: most recent test-runner evidence exit_code=1, contradicting the claim
  verifier:  test_result_verifier_v1
  when:      2026-08-29T03:14:07Z

🛡 ✓ VERIFIED
  claim:     The build succeeded
  rationale: most recent build evidence exit_code=0, matching the claim
  verifier:  test_result_verifier_v1
  when:      2026-08-29T02:58:41Z

If nothing has been recorded yet:

No findings recorded yet.

If the daemon isn't reachable:

fornax: daemon unreachable (is `fornax-daemon` running?)

What each field means

Field	Meaning
verdict (icon + word)	One of the five states — see Claims, Evidence & Findings.
claim	The literal claim text extracted from the agent's transcript.
rationale	Why the verifier reached this verdict — always present, never blank, so a CONTRADICTED or UNAVAILABLE finding is actionable rather than just a color.
verifier	The verifier's stable name (e.g. test_result_verifier_v1), recorded on every finding it produces — useful once more than one verifier exists.
when	UTC timestamp the finding was computed.

Wiring it into a custom keybinding or command

fornax detail is a plain CLI call reading from the daemon's localhost HTTP API (GET /api/findings/recent on FORNAX_HTTP_PORT, default 4317) — you can wire it into an editor command, a shell alias, or a Claude Code custom slash command exactly like any other local tool. There's nothing Fornax-specific about the wiring itself; see Configuration for the two environment variables that matter (FORNAX_HOME, FORNAX_HTTP_PORT).