@chrono-meta/fh-gate 1.4.14 → 1.4.16

package/CLAUDE.md

@@ -103,17 +103,28 @@ Simplification guard: trivial denials with one obvious fix → state block + sin
 
 **4-step summary**: ① Auto-read CLAUDE.md + CATALOG + session card + registry scan → ② One-line proposal (new user / exploratory / returning branches) → ③ 5-skill cascade (plugin-recommender → synergy → .claudeignore → model → verify) → ④ Approval + setup
 
-**Returning-user door skeleton (summary-level — applies even if the detail file read is skipped; returning users only, new/exploratory branches stay in the detail file)**: open with the fixed 3-door menu, **🐿️ on its own line as the skeleton's first line** —
+**Greeting branch + door skeleton (summary-level — applies even if the detail file read is skipped)**: the branch test is **mechanical local state — session files under `tracks/`** — never git log / CATALOG residue (a fresh clone carries full history but zero session files: it is a NEW install — origin: a fresh-clone sonnet sim rendered the returning menu off commit messages, `fh_signal_2026-06-11` FP8). Every variant opens with **🐿️ on its own line as the skeleton's first line** — the marker is part of the skeleton, one salience unit with the menu, not a separate rule.
 
-> 🐿️
-> *"① Connect/map a project · ② Work on a mapped project — {field candidates} · ③ FH self-development — {FH worklist}"*
+- **New user** (no session files AND no mapped project tracks under `tracks/` — fresh clone/install; underscore meta dirs `_meta`/`_audit`/`_contrib` don't count): 2-door starter, never the returning menu —
 
-The marker is part of this skeleton — one salience unit with the menu, not a separate rule (origin: a sonnet-tier sim emitted the menu but dropped the standalone marker rule, `fh_signal_2026-06-11`). Compose session-card candidates **into doors ②/③**, never as a raw priority dump that replaces the menu. An urgent open item (time-windowed handoff · blocking external deadline) outranks the menu; an explicit task utterance skips it entirely (see Guards below); cadence reminders (§Cadence Rules) ride below it, they don't displace it. Canonical source: `fh_detail_protocols.md` Step 2 §Returning user — keep door labels in sync.
+  > 🐿️
+  > *"Looks like you're new here! ① Create your first project (guided) · ② Map an existing project — and I can run `/install-wizard` to finish initial setup."*
 
-**Identity marker**: every greeting response (Step ②) opens with 🐿️ on its own line. For returning users it is embedded in the door skeleton above (do not strip it when composing doors); new/exploratory branch templates carry it in `fh_detail_protocols.md` Step 2.
+- **Returning user** (session files OR mapped project tracks exist): fixed 4-door menu —
+
+  > 🐿️
+  > *"① Map a project · ② Create a new project · ③ Accelerate a mapped project (work · Full-Harness · skills/agents/plugins) — {field candidates} · ④ Cross-project synergy"*
+
+  Render conditions: ①②③ always (③'s candidates composed live) · ④ only when **2+ project tracks** exist (underscore meta dirs don't count) — synergy findings flow back into each project, and may *propose* an FH contribution (`/field-harvest` → `tracks/_contrib`) as an **outcome of findings, never a standing door**.
+
+- **Developer door (unnumbered, outside the menu)**: when **FH-dev state exists** (session card `tracks/_meta/reference_next_session_starter.md` · open `fh_signal_*` files · `CLAUDE.local.md`), append to the menu line: ` · 🔧 FH self-development — {FH worklist}`. The hub operator always has this state, so the owner always sees it — no flag needed. Without dev state the door is **silently absent**; the user typing `developer` / `개발자` **as a standalone utterance or menu reply** (not a substring of a task sentence) opens it on demand (routes to `docs/CONTRIBUTING.md` + `tracks/_contrib/` + open `fh_signal_*` items).
+
+Compose session-card candidates **into door ③ (field) and the 🔧 door (FH-dev)**, never as a raw priority dump that replaces the menu. An urgent open item (time-windowed handoff · blocking external deadline) outranks the menu; an explicit task utterance skips it entirely (see Guards below); cadence reminders (§Cadence Rules) ride below it, they don't displace it. Canonical source: `fh_detail_protocols.md` Step 2 — keep branch tests and door labels in sync.
+
+**Identity marker**: every greeting response (Step ②) opens with 🐿️ on its own line. It is embedded in both skeletons above (do not strip it when composing doors); the exploratory branch template carries it in `fh_detail_protocols.md` Step 2.
 
 **Guards**: explicit task-entry utterance → skip onboarding · once per session · code/debug requests → start working directly · project routing is a suggestion, mention at most once
-**Metadata-is-not-intent guard**: the trigger is the user's **typed message only**. Session metadata — branch name (auto-derived from the first message, e.g. `claude/korean-greeting-*`), repo name, file paths — is **never** a task spec and never suppresses or redirects the greeting trigger. A bare greeting fires onboarding even when the branch name looks like a feature request; if the only "task" signal lives in metadata and not in what the user typed, treat the message as a greeting and run the 3-axis scaffold.
+**Metadata-is-not-intent guard**: the trigger is the user's **typed message only**. Session metadata — branch name (auto-derived from the first message, e.g. `claude/korean-greeting-*`), repo name, file paths — is **never** a task spec and never suppresses or redirects the greeting trigger. A bare greeting fires onboarding even when the branch name looks like a feature request; if the only "task" signal lives in metadata and not in what the user typed, treat the message as a greeting and run the greeting branch + door skeleton above.
 
 ## New Skill Creation Pre-Commit Gate
 
@@ -183,7 +194,11 @@ the operator (one line), mirroring §Floor governance.
 
 If `model:`-pinned dispatch is unavailable (plan/billing gate), fall back to a cross-session headless
 run (`claude -p "<trigger>" --model <tier>` in the target cwd) — stronger isolation, zero instruction
-contamination. 2026-06-15+: headless `claude -p` draws from the hard-capped credit pool, not the
+contamination. **Saturation disguise (N=2, 2026-06-11/12)**: the same "Usage credits required for 1M
+context" error also fires when the *session* is near context saturation, not the plan gate — in a
+long-running session, compact (flush handoff state to disk first) and retry the dispatch once before
+concluding the gate is closed (identical opus-pinned dispatch failed pre-compaction, succeeded
+post-compaction 2026-06-12). 2026-06-15+: headless `claude -p` draws from the hard-capped credit pool, not the
 subscription — prefer in-session Agent dispatch when the plan gate allows; take the headless fallback
 knowingly. Record sim results in the Axes 2–3 marker + sub-agent invocation log.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@chrono-meta/fh-gate",
-  "version": "1.4.14",
+  "version": "1.4.16",
   "description": "FH runtime adapters — run FH governance, skills, and agents via Claude or Codex with machine-parseable gates.",
   "license": "MIT",
   "keywords": [

package/plugins/fh-meta/agents/expert.md

@@ -24,10 +24,10 @@ What expert grounds against:
 What expert does NOT rely on:
   - Its own unverified parametric recall ("I think X is true")
   - Internal hub assets only (that is fact-checker's job — internal grep)
-  - Declared-source-file back-tracing only (that is source-grounding-audit's job)
+  - Declared-source-file back-tracing only (that is phantom-quench's job)
 ```
 
-**Boundary (no overlap)**: `fact-checker` greps the *hub/own environment*; `source-grounding-audit` traces claims to *declared internal source files*; `expert` checks against the *external world / frontier*. The three cover internal-duplication, internal-provenance, and external-accuracy respectively.
+**Boundary (no overlap)**: `fact-checker` greps the *hub/own environment*; `phantom-quench` traces claims to *declared internal source files*; `expert` checks against the *external world / frontier*. The three cover internal-duplication, internal-provenance, and external-accuracy respectively.
 
 ## Accuracy Matrix
 

package/plugins/fh-meta/skills/goal-quench/SKILL.md

@@ -296,6 +296,14 @@ Output to user (one line only):
 
 Do not ask for confirmation. The user may override by re-running `/goal-quench --sidecar none`.
 
+**Dispatch-failure triage (saturation disguise)**: a sidecar dispatch that fails with a 1M-context /
+usage-credits error late in a long run may be reporting *session context saturation*, not a billing
+gate (measured 2026-06-12: identical error pre-compaction; the same opus-pinned dispatch succeeded
+post-compaction with the sub-agent completing normally). Triage order: compact — flushing handoff
+state to disk first — → retry the dispatch once → only then take the headless `claude -p` fallback
+(credit-pool cost) or an inline at-floor pass. Concluding "billing gate" from the first failure
+skips the cheapest recovery.
+
 ### Hand-off
 
 After orchestration, **update** (not re-create) `.claude/goal-quench.active` — add `mode:` and `composed_plan:` lines alongside existing budget fields. Re-creating the file loses the `start_commit` field written in Phase 1 Step 3. Then proceed to threshold injection (Phase 1 Step 4) and hand off to /goal. Phase 3 verification then runs as in core — `pipeline-conductor --quick` for pro, `--full` for max (max implies external-facing stakes).

package/plugins/fh-meta/skills/phantom-quench/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: phantom-quench
-description: The grounding member of the quench series — extracts proper nouns, numerical values, and branching conditions from artifacts (TCs, analysis reports, design documents), back-traces them to declared source files, and marks anything not found as a Phantom Claim (ungrounded — present in the artifact but not traceable to a declared source; not a claim that it is necessarily false). If steel-quench attacks output patterns (self-declarations, cushion language), phantom-quench attacks input tracing (where did this come from?). Renamed from source-grounding-audit (2026-06-06, quench-series); `/source-grounding-audit` still resolves as an alias. Triggered by "phantom detection", "phantom-quench", "phantom claim", "hallucinated claim detection", "source back-trace", "source audit", "verify source", "TC evidence tracing", "where did this come from", "grounding audit", "source grounding audit", "false claim detection".
+description: The grounding member of the quench series — extracts proper nouns, numerical values, and branching conditions from artifacts (TCs, analysis reports, design documents), back-traces them to declared source files, and marks anything not found as a Phantom Claim (ungrounded — present in the artifact but not traceable to a declared source; not a claim that it is necessarily false). If steel-quench attacks output patterns (self-declarations, cushion language), phantom-quench attacks input tracing (where did this come from?). Renamed from source-grounding-audit (2026-06-06, quench-series); the old name appears here so legacy references still route to this skill (alias stub directory removed 2026-06-12). Triggered by "phantom detection", "phantom-quench", "phantom claim", "hallucinated claim detection", "source back-trace", "source audit", "verify source", "TC evidence tracing", "where did this come from", "grounding audit", "source grounding audit", "false claim detection".
 user-invocable: true
 allowed-tools: ["Read", "Write", "Edit", "Bash", "Grep", "Glob"]
 model: sonnet
@@ -13,8 +13,10 @@ model: sonnet
 > **Renamed from `source-grounding-audit` (2026-06-06)** — the grounding member of the quench series
 > (steel-quench · phantom-quench · goal-quench). Same skill, same ruleset; only the label changed to fit
 > the family. The **v1 paper (Zenodo 10.5281/zenodo.20397566) cites the old name** — that is the
-> historical record, not a phantom. `/source-grounding-audit` still resolves via the deprecated redirect
-> stub at `plugins/fh-meta/skills/source-grounding-audit/SKILL.md` (`successor: phantom-quench`).
+> historical record, not a phantom (`paper/forge_harness_v1.0.html` is left unchanged by design; readers
+> map *source-grounding-audit (v1) = phantom-quench (current)*). The deprecated redirect stub directory
+> was removed 2026-06-12 — old-name utterances ("source grounding audit", "grounding audit") route here
+> via this description's trigger phrases.
 > This is a **label rename, not a capability change** — phantom-quench does not fuse steel-quench or
 > inject faults; those remain separate (orthogonality is deliberate — see Role Separation below).
 >

package/scripts/selfcheck.sh

@@ -65,6 +65,27 @@ count_check "marketplace.json fh-meta" .claude-plugin/marketplace.json
 count_check "README header"            README.md                                     "${total_sk} skills · ${total_ag} agents"
 count_check "local_fh_context fh-meta" templates/local_fh_context.md                 "(fh-meta, ${meta_sk})"
 
+# Referenced-path existence: backtick-quoted repo-relative file refs in the always-loaded
+# governance surface (CLAUDE.md + .claude/rules/*.md) must exist. Phantom-reference class
+# recurred N>=3 in the 2026-06-11 audit window (operations.md _scanner.sh, claude-chrono path,
+# stale templates ref) — instrument-not-habit. Globs/placeholders/{vars} are excluded by the
+# filter; tracks/ is machine-local and deliberately out of scope. Gitignored refs (e.g.
+# `.claude/settings.json` named in prose *about* gitignored files) are skipped — they exist
+# locally but not on a fresh clone, and "must exist" here means "must ship".
+while IFS= read -r p; do
+  if git check-ignore -q "$p" 2>/dev/null; then
+    echo "SKIP  ref-path (gitignored): $p"
+  elif [ -f "$p" ]; then
+    echo "PASS  ref-path: $p"
+  else
+    echo "FAIL  ref-path: $p — referenced in CLAUDE.md/.claude/rules but missing"
+    fail=1
+  fi
+done < <(grep -hoE '\`[^\` ]+\`' CLAUDE.md .claude/rules/*.md 2>/dev/null \
+  | sed 's/\`//g' \
+  | grep -E '^(knowledge|templates|scripts|docs|plugins|\.claude)/[^*{}<>$]+\.(md|sh|ya?ml|jsonc|json)$' \
+  | sort -u)
+
 if [ "$fail" -ne 0 ]; then
   echo "SELFCHECK: FAIL"
   exit 1

package/plugins/fh-meta/skills/source-grounding-audit/SKILL.md

@@ -1,42 +0,0 @@
----
-name: source-grounding-audit
-description: >-
-  RENAMED to phantom-quench (2026-06-06, quench-series rebrand). Same skill, same ruleset — only the
-  label changed to fit the quench family (steel-quench · phantom-quench · goal-quench). Use
-  /phantom-quench. This alias is retained so old references and the v1 paper's name still resolve.
-user-invocable: false
-allowed-tools: []
-model: sonnet
-deprecated: true
-deprecated_reason: renamed to phantom-quench (label-only; not a merge — same skill)
-deprecated_date: 2026-06-06
-successor: phantom-quench
----
-
-# source-grounding-audit — RENAMED to `phantom-quench`
-
-> **Renamed to `phantom-quench` (2026-06-06).** This is a **label rename, not a deprecation-by-merge** —
-> the skill is unchanged and fully active under its new name. Invoke **`/phantom-quench`**.
-
-## Why the rename
-
-phantom-quench is the **grounding member of the quench series** (steel-quench attacks output patterns ·
-phantom-quench traces inputs for Phantom Claims · goal-quench gates autonomous runs). The old descriptive
-name did not signal that family membership; the function is identical.
-
-## Where the skill lives now
-
-`plugins/fh-meta/skills/phantom-quench/SKILL.md` (+ `SKILL_detail.md`) — full ruleset preserved
-(S-grade blocker, Human Gate, Pattern Diagnosis, etc.).
-
-## Record note (do not "fix")
-
-The **v1 paper** (Zenodo 10.5281/zenodo.20397566; arXiv submission) cites `source-grounding-audit`.
-That is the **immutable historical name**, not a phantom — `paper/forge_harness_v1.0.html` is left
-unchanged by design. Future readers map: *source-grounding-audit (v1 paper) = phantom-quench (current)*.
-
-## Done When
-
-Deprecated alias — no active execution path of its own. Done When: all invocation routes through
-`/phantom-quench` (the successor); this entry exists only so old names resolve. Satisfies the
-harness-doctor L2 M-tier Done-When requirement (CLAUDE.md §New Skill Creation Pre-Commit Gate).