Trajectory step 3: tool.call events from native session logs

CHANGELOG.md

@@ -6,6 +6,7 @@ Versions follow [SemVer](https://semver.org).
 
 ## [Unreleased]
 
+- Session end records each tool call from the retained Claude or Codex log as a `tool.call` trajectory event, once per call even when a session resumes. `outerloop trajectory` and a new Process section in the PR body show counts by tool. Older kernels ignore these events.
 - `outerloop status` shows a run's priced session cost as a lower bound when some legs are unpriced (a wake killed mid-turn leaves no usage event), for example `session-cost>=$4.2000 (1 unpriced)`, instead of hiding the priced part behind `unknown`; the status row and JSON gain the priced-leg count (#516).
 - The pre-PR panel accepts a composite change — a PR that carries several changes — when the report gives each component's isolated measurement and the combination's under the same protocol and the combination is what the kernel measured. A bundle without those isolated measurements stays blocked (#489).
 - Claude sessions can use a subscription token from `claude setup-token` (`sk-ant-oat...`). The harness sends it as `CLAUDE_CODE_OAUTH_TOKEN`, not `ANTHROPIC_API_KEY`, and `init` accepts it. A panel judge on a token runs `claude --safe-mode`, because `--bare` cannot read OAuth tokens. Both modes ignore the checkout's CLAUDE.md, hooks, plugins and MCP servers. The pinned Claude Code release (2.1.285) has `--safe-mode`.

docs/design/trajectory.md

@@ -46,4 +46,10 @@ directory. Self-reported measures record the ruler alone: they have no
 evaluation image or result files. A pending submission does not yet name the
 trajectory head: kernels before this one reject unknown pending fields, so the
 pending reader now ignores them, and a later release can add the field without
-breaking this one. Tool calls and attestation are later steps.
+breaking this one. Session end also appends one `tool.call` event per tool call
+found in the retained, redacted native log: Claude project transcripts and Codex
+rollout JSONL are parsed host-side, bounded to 500 calls per session, and Hermes
+or unknown formats yield nothing. Each event carries the tool name, repo-relative
+files, a bounded shell command and, when the log has them, the exit code and
+outcome — never file contents or tool output. `outerloop trajectory` counts them
+by tool and the PR body summarizes the top tools. Attestation is a later step.

src/outerloop/attempt.py

@@ -152,12 +152,22 @@
 from outerloop.syscall import refresh_tool as syscall_refresh_tool
 from outerloop.syscall import write_siblings as syscall_write_siblings
 from outerloop.trajectory import file_refs
+from outerloop.trajectory import read as read_trajectory
 from outerloop.trajectory import record as record_event
+from outerloop.trajectory import summarize as summarize_trajectory
 from outerloop.verifier import MAX_CLAIM_CHARS
 
 REPLY_MARKER = marker("followup")
 
 
+def tool_call_counts(run_dir: Path) -> dict[str, int]:
+    """The run's recorded tool calls by tool; an unreadable chain reads as none."""
+    try:
+        return summarize_trajectory(read_trajectory(run_dir))["tool_calls_by_tool"]
+    except OSError:
+        return {}
+
+
 def _pr_number(pr_url: str) -> int:
     tail = pr_url.rstrip("/").rsplit("/", 1)[-1]
     if not tail.isdigit():
@@ -4724,6 +4734,7 @@ def refuse(
                 display_digits=bench.display_digits,
                 experiments=experiments_rows(run_dir),
                 base_sha=base_sha,
+                tool_calls=tool_call_counts(run_dir),
             )
             body += f"\n\n{progress_link(config.target, github)}\n"
             if issue_number:

src/outerloop/cli.py

@@ -1360,9 +1360,16 @@ def main(argv: list[str] | None = None) -> int:
                     f"{verb} {count}" for verb, count in summary["refusals_by_verb"].items()
                 )
                 authors = ", ".join(summary["authors"]) or "unknown"
+                tools = ", ".join(
+                    f"{tool} {count}"
+                    for tool, count in sorted(
+                        summary["tool_calls_by_tool"].items(), key=lambda item: (-item[1], item[0])
+                    )
+                )
                 print(
                     f"refusals: {summary['refusals']} of {summary['decisions']} decisions"
-                    f"{f' ({verbs})' if verbs else ''}; author: {authors}"
+                    f"{f' ({verbs})' if verbs else ''}; author: {authors}; "
+                    f"tool calls: {summary['tool_calls']}{f' ({tools})' if tools else ''}"
                 )
         return 1 if problems and problems != ["no trajectory"] else 0
     if args.command == "status":

src/outerloop/orchestrator.py

@@ -2821,6 +2821,20 @@ def experiments_section(rows: list[dict[str, Any]]) -> list[str]:
     return lines
 
 
+MAX_PROCESS_TOOLS = 5
+
+
+def process_section(tool_calls: dict[str, int] | None) -> list[str]:
+    """Total and top tools from the run's trajectory; empty when there are none."""
+    total = sum((tool_calls or {}).values())
+    if not total:
+        return []
+    ranked = sorted((tool_calls or {}).items(), key=lambda item: (-item[1], item[0]))
+    top = ", ".join(f"{name} {count}" for name, count in ranked[:MAX_PROCESS_TOOLS])
+    calls = "call" if total == 1 else "calls"
+    return ["", "## Process", "", f"{total} tool {calls} from the run's trajectory: {top}."]
+
+
 def pr_body(
     result: AttemptResult,
     config: RunConfig,
@@ -2829,6 +2843,7 @@ def pr_body(
     experiments: list[dict[str, Any]] | None = None,
     *,
     base_sha: str = "",
+    tool_calls: dict[str, int] | None = None,
 ) -> str:
     """The PR body for an improved run: the author's report, the experiments
     the run actually ran (from the launch ledger), the measured table, and
@@ -2909,6 +2924,7 @@ def pr_body(
             "",
             *report_lines,
             *experiments_section(experiments or []),
+            *process_section(tool_calls),
             "",
             "## Measured",
             "",
@@ -2940,6 +2956,7 @@ def pr_body(
                 "",
                 redact(result.submit_report, redact_secrets)[:MAX_REPORT_BODY],
                 *experiments_section(experiments or []),
+                *process_section(tool_calls),
                 "",
                 "## Self-reported values",
                 "",

src/outerloop/session_evidence.py

@@ -23,6 +23,7 @@
 from pathlib import Path
 from typing import TYPE_CHECKING, Any
 
+from outerloop.tool_calls import record_tool_calls
 from outerloop.trajectory import session_event
 
 if TYPE_CHECKING:
@@ -320,6 +321,10 @@ def wrapped(
                     )
                 except Exception:
                     log.warning("could not retain session evidence", exc_info=True)
+                try:
+                    record_tool_calls(workspace, backend, result.session_record_path)
+                except Exception:
+                    log.warning("could not record session tool calls", exc_info=True)
                 session_event(
                     workspace, "session.end", backend, self.model, result.session_id, result=result
                 )