Fix cross-repo gating: absolute paths for finalize_report, skip gate for non-toolkit repos

Skills used relative path for finalize_report.py which failed when CWD was a different project. Gate hook now skips enforcement when no gates.json exists, so non-toolkit repos are not blocked.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: jvalin17

hooks/gate_hook.py

@@ -68,7 +68,7 @@ def load_gate_config(project_dir: Path) -> dict:
                 return json.loads(candidate.read_text(encoding="utf-8"))
             except (json.JSONDecodeError, OSError):
                 pass
-    return {"enforcement": "block", "gate_mode": "legacy"}
+    return {"enforcement": "block", "gate_mode": "legacy", "_no_config": True}
 
 
 def get_config_value(config: dict, key: str, default=None):
@@ -325,6 +325,8 @@ def run_gate(
         return 0, ""
 
     config = load_gate_config(project_dir)
+    if config.get("_no_config"):
+        return 0, ""  # No gates.json — not a toolkit-managed project
     gate_mode = config.get("gate_mode", "legacy")
     enforcement = resolve_enforcement(
         config.get("enforcement", "block"),

skills/assess/SKILL.md

@@ -187,7 +187,7 @@ passes only when `fix_now` is 0 and mechanical test/lint re-runs pass.
 Then run:
 
 ```
-python3 hooks/finalize_report.py assess .scratch/assess_<slug>/findings.json
+python3 /Users/jvalin/dev/st5/agent-toolkit/hooks/finalize_report.py assess .scratch/assess_<slug>/findings.json
 ```
 
 The hook writes `reports/assess/assess_<slug>_<id>.md` and prints a JSON

skills/evaluate/SKILL.md

@@ -161,7 +161,7 @@ score than your dimension scores support.
 Then run:
 
 ```
-python3 hooks/finalize_report.py evaluate .scratch/evaluate_<slug>/findings.json
+python3 /Users/jvalin/dev/st5/agent-toolkit/hooks/finalize_report.py evaluate .scratch/evaluate_<slug>/findings.json
 ```
 
 The hook re-runs `test_command` and `lint_command` from `gates.json` itself —

skills/precommit/SKILL.md

@@ -125,7 +125,7 @@ Findings schema (all keys required):
 Then run:
 
 ```
-python3 hooks/finalize_report.py precommit .scratch/precommit_<slug>/findings.json
+python3 /Users/jvalin/dev/st5/agent-toolkit/hooks/finalize_report.py precommit .scratch/precommit_<slug>/findings.json
 ```
 
 The hook re-runs `test_command` and `lint_command` from `gates.json` itself —

skills/reviewer/SKILL.md

@@ -104,7 +104,7 @@ only when `high` is 0 and mechanical test/lint re-runs pass.
 Then run:
 
 ```
-python3 hooks/finalize_report.py reviewer .scratch/reviewer_<slug>/findings.json
+python3 /Users/jvalin/dev/st5/agent-toolkit/hooks/finalize_report.py reviewer .scratch/reviewer_<slug>/findings.json
 ```
 
 The hook writes `reports/reviewer/review_<slug>_<id>.md` and prints a JSON