"""Prompt-size diagnostic: ``hermes prompt-size``. Builds a real inspection agent (so the numbers match what ships on the wire) but never makes a network call: dummy credentials force ``AIAgent.__init__`` down the direct-construction path, then ``build_system_prompt_parts`` / ``agent.tools`` are inspected offline. """ from __future__ import annotations import json import re from pathlib import Path from typing import Any, Dict, List, Optional, Tuple _SKILLS_BLOCK_RE = re.compile(r".*?", re.DOTALL) # A rendered skill entry is ``- name: desc`` (or ``- name``) indented four spaces under its category # header, or two when it is a one-skill category's only row; headers never start with ``- ``. _SKILL_LINE_PREFIX = "- " # Posture-demoted categories render all visible skill names on one shared line. _NAMES_ONLY_LINE_RE = re.compile(r"^ .+ \[names only\]: (?P.+)$") # Cap the human-readable "Skills by size" table; ``--json`` always has them all. _SKILLS_TABLE_LIMIT = 20 def _bytes(s: str) -> int: return len(s.encode("utf-8")) def _size(text: str) -> dict[str, int]: return {"chars": len(text), "bytes": _bytes(text)} def _fmt_kb(n: int) -> str: return f"{n / 1024:.1f} KB" def _tool_name(tool: Any) -> str: """Callable name of a tool schema (OpenAI ``function`` shape).""" if not isinstance(tool, dict): return "" fn = tool.get("function") return str(fn["name"]) if isinstance(fn, dict) and fn.get("name") else str(tool.get("name", "")) def _build_inspection_agent(platform: str) -> Any: """Offline AIAgent for prompt inspection: dummy ``api_key`` + ``base_url`` force the direct-construction path (no provider auto-detection, no network); toolsets resolve the way the gateway does so the breakdown matches a real session. """ from run_agent import AIAgent from hermes_cli.config import load_config from hermes_cli.tools_config import _get_platform_tools from agent.skill_utils import parse_config_string_list cfg = load_config() model_cfg = cfg.get("model", {}) if isinstance(cfg.get("model"), dict) else {} agent_cfg = cfg.get("agent") or {} return AIAgent( model=model_cfg.get("default") or model_cfg.get("model") or "", api_key="inspect-only", base_url="https://openrouter.ai/api/v1", quiet_mode=True, save_trajectories=False, platform=platform, enabled_toolsets=sorted(_get_platform_tools(cfg, platform)), disabled_toolsets=parse_config_string_list(agent_cfg.get("disabled_toolsets")) or None, ) def _skill_md_paths_by_name() -> dict[str, Path]: """Map each installed skill's frontmatter ``name`` AND directory name to its ``SKILL.md``. Local skills win over external dirs (``get_all_skills_dirs`` yields local first), matching the index's own precedence. """ from agent.skill_utils import get_all_skills_dirs, iter_skill_index_files, parse_frontmatter mapping: dict[str, Path] = {} for skills_dir in get_all_skills_dirs(): if not skills_dir.exists(): continue for skill_file in iter_skill_index_files(skills_dir, "SKILL.md"): dir_name = skill_file.parent.name try: frontmatter, _ = parse_frontmatter( skill_file.read_text(encoding="utf-8-sig") ) frontmatter_name = str(frontmatter.get("name") or dir_name) except Exception: frontmatter_name = dir_name mapping.setdefault(frontmatter_name, skill_file) # first (local) occurrence wins mapping.setdefault(dir_name, skill_file) return mapping def _compute_skills_breakdown(skills_block: str) -> list[dict[str, Any]]: """Per-skill byte breakdown parsed from the rendered ````. ``index_line_bytes`` is the skill's attributed always-on index cost. For a compact ``[names only]`` line each name keeps its own bytes plus an even share of the shared prefix and separators. """ name_to_path = _skill_md_paths_by_name() entries: list[dict[str, Any]] = [] def append_entry(name: str, **index_fields: int) -> None: # kwarg order == output key order path = name_to_path.get(name) md_bytes: Optional[int] = None try: md_bytes = path.stat().st_size if path is not None else None except OSError: pass entries.append({"name": name, **index_fields, "skill_md_bytes": md_bytes, "path": str(path) if path is not None else ""}) for line in skills_block.splitlines(): line_bytes = _bytes(line) if (compact_match := _NAMES_ONLY_LINE_RE.match(line)) is not None: names = [n.strip() for n in compact_match.group("names").split(",") if n.strip()] name_bytes = [_bytes(name) for name in names] shared_base, shared_remainder = divmod(line_bytes - sum(name_bytes), len(names)) if names else (0, 0) for index, name in enumerate(names): shared = shared_base + (1 if index < shared_remainder else 0) append_entry(name, index_line_bytes=name_bytes[index] + shared, index_line_total_bytes=line_bytes, index_line_shared_bytes=shared, index_line_skill_count=len(names)) elif (row := line.lstrip()).startswith(_SKILL_LINE_PREFIX): # Partition on ``": "`` (not ``:``) so namespaced names like ``codex:rescue`` stay intact. name = row[len(_SKILL_LINE_PREFIX):].partition(": ")[0].strip() if name: append_entry(name, index_line_bytes=line_bytes, index_line_total_bytes=line_bytes, index_line_shared_bytes=0, index_line_skill_count=1) entries.sort(key=lambda e: (-(e["skill_md_bytes"] or 0), e["name"])) return entries def _compute_toolsets_breakdown(tools: list[Any]) -> list[dict[str, Any]]: """Per-toolset schema-byte breakdown, largest-first (tie-broken by name). Each tool is attributed to its single canonical toolset so ``json_bytes`` sums to the grand total. """ from tools.registry import registry tool_to_toolset = registry.get_tool_to_toolset_map() groups: dict[str, dict[str, Any]] = {} for tool in tools: toolset = tool_to_toolset.get(_tool_name(tool)) or "(unknown)" group = groups.setdefault(toolset, {"toolset": toolset, "tool_count": 0, "json_bytes": 0}) group["tool_count"] += 1 group["json_bytes"] += _bytes(json.dumps(tool, ensure_ascii=False)) return sorted(groups.values(), key=lambda g: (-g["json_bytes"], g["toolset"])) def compute_prompt_breakdown(platform: str = "cli") -> dict[str, Any]: """Prompt-size measurements for a fresh session: ``system_prompt``, ``skills_index``, ``memory``, ``user_profile``, ``tools``, ``sections`` (the three prompt tiers), and the largest-first ``skills_breakdown`` / ``toolsets_breakdown`` ("what should I disable?"). """ from agent.system_prompt import build_system_prompt, build_system_prompt_parts agent = _build_inspection_agent(platform) parts = build_system_prompt_parts(agent) full = build_system_prompt(agent) stable, context, volatile = (parts.get(k, "") for k in ("stable", "context", "volatile")) # The skills index lives in the volatile tier (moved from stable so skill edits don't # invalidate the cached identity prefix); fall back to stable for older layouts. skills_match = _SKILLS_BLOCK_RE.search(volatile) or _SKILLS_BLOCK_RE.search(stable) skills_index = skills_match.group(0) if skills_match else "" # Memory + user profile are joined into ``volatile``; re-derive them from the store so the # numbers stay attributable. memory_block = user_block = "" store = getattr(agent, "_memory_store", None) if store is not None: try: if getattr(agent, "_memory_enabled", True): memory_block = store.format_for_system_prompt("memory") or "" if getattr(agent, "_user_profile_enabled", True): user_block = store.format_for_system_prompt("user") or "" except Exception: pass tools = getattr(agent, "tools", None) or [] sections: list[tuple[str, int, int]] = [ (label, len(text), _bytes(text)) for label, text in (("stable (identity/guidance/skills)", stable), ("context (AGENTS.md/cwd files)", context), ("volatile (memory/profile/timestamp)", volatile)) ] return { "platform": platform, "model": getattr(agent, "model", "") or "", "system_prompt": _size(full), "skills_index": _size(skills_index), "memory": _size(memory_block), "user_profile": _size(user_block), "tools": {"count": len(tools), "json_bytes": _bytes(json.dumps(tools, ensure_ascii=False))}, "sections": sections, "skills_breakdown": _compute_skills_breakdown(skills_index), "toolsets_breakdown": _compute_toolsets_breakdown(tools), } def render_breakdown(data: dict[str, Any]) -> str: """Render the breakdown as plain text suitable for a terminal.""" sp = data["system_prompt"] tools = data["tools"] lines: list[str] = [ f"Prompt-size breakdown (platform={data['platform']}, model={data['model'] or 'unset'})", "", f" System prompt total : {sp['bytes']:>8,} B ({_fmt_kb(sp['bytes'])}, {sp['chars']:,} chars)", "", " Major blocks:", ] for label, key in (("skills index", "skills_index"), ("memory", "memory"), ("user profile", "user_profile")): byts = data[key]["bytes"] lines.append(f" {label:<19}: {byts:>8,} B ({_fmt_kb(byts)})") lines += ["", " Prompt tiers:"] + [f" {label:<36}: {byts:>8,} B ({_fmt_kb(byts)})" for label, _chars, byts in data["sections"]] lines += ["", f" Tool schemas : {tools['json_bytes']:>8,} B ({_fmt_kb(tools['json_bytes'])}, {tools['count']} tools)"] if toolsets := data.get("toolsets_breakdown") or []: lines += ["", " Toolsets by size (tool-schema JSON, largest first):", f" {'toolset':<22} {'tools':>5} {'schema':>10}"] lines += [f" {ts['toolset']:<22} {ts['tool_count']:>5} {ts['json_bytes']:>8,} B ({_fmt_kb(ts['json_bytes'])})" for ts in toolsets] # Per-skill cost — index line (always shipped) vs SKILL.md (read on load). if skills := data.get("skills_breakdown") or []: lines += ["", " Skills by size (SKILL.md on-disk = read cost; index cost = attributed always-on bytes, largest first):", f" {'skill':<28} {'SKILL.md':>10} {'index cost':>10}"] shown = skills[:_SKILLS_TABLE_LIMIT] for sk in shown: md = sk["skill_md_bytes"] md_str = f"{md:>8,} B" if md is not None else f"{'n/a':>10}" name = sk["name"] if len(sk["name"]) <= 28 else sk["name"][:27] + "…" lines.append(f" {name:<28} {md_str} {sk['index_line_bytes']:>8,} B") if (remaining := len(skills) - len(shown)) > 0: lines.append(f" … and {remaining} more (use --json for the full list)") return "\n".join(lines) def cmd_prompt_size(args: Any) -> None: """Entry point for ``hermes prompt-size``.""" try: data = compute_prompt_breakdown(getattr(args, "platform", "cli") or "cli") except Exception as e: print(f"Could not compute prompt-size breakdown: {e}") return print(json.dumps(data, ensure_ascii=False, indent=2) if getattr(args, "json", False) else render_breakdown(data))