All plugins

aux-ledger

Local ledger of Hermes's auxiliary LLM calls (titling, compression, vision, approval): /aux shows tokens, time and failures per task, model, provider or day, or lists the newest calls (ten, or up to 100 with /aux recent 50), for the last 24 hours, any number of hours or days, the latest session or everything recorded. Disclosure — writes auxiliary-call metadata (ids, models, timings, token counts, no prompt or response text) to plugin-data/aux-ledger/calls.jsonl, trimmed to 5,000 rows past 2 MB; no network

olric3015Toolsv1.4.0Updated 1 month ago

About

Local ledger of Hermes's auxiliary LLM calls (titling, compression, vision, approval): /aux shows tokens, time and failures per task, model, provider or day, or lists the newest calls (ten, or up to 100 with /aux recent 50), for the last 24 hours, any number of hours or days, the latest session or everything recorded. Disclosure — writes auxiliary-call metadata (ids, models, timings, token counts, no prompt or response text) to plugin-data/aux-ledger/calls.jsonl, trimmed to 5,000 rows past 2 MB; no network

aux-ledger

A local ledger of the LLM calls Hermes makes behind a turn, and a /aux command that shows what
they cost.

Hermes calls a model for more than your turns: session titles, context compression, vision,
approval checks and other auxiliary tasks each make their own requests, with retries and
fallbacks. Those requests do not pass through the main-loop pre_api_request /
post_api_request hooks, so usage plugins built on them do not count this traffic. aux-ledger
subscribes to post_auxiliary_call and records one row per provider attempt.

Install

hermes plugins install aux-ledger --enable

Requires Hermes 0.21.5 or newer (the first release with the auxiliary-call hooks). No API key,
no configuration, no dependencies.

Use

/aux             last 24 hours, by task
/aux session     the most recent session that made an auxiliary call
/aux all         everything in the ledger
/aux 7d          the last 7 days (any number of hours or days: 6h, 30d)
/aux models      add to any of the above to group by model instead of task
/aux providers   add to any of the above to group by provider instead of task
/aux days        add to any of the above to group by day, newest first
/aux recent      add to any of the above to list the newest 10 calls instead
/aux recent 25   the newest 25 instead (1 to 100)
/aux clear       delete the ledger

A time window counts back from now, so /aux 7d answers "what did the background calls cost
this week" and /aux 7d providers splits that bill by provider. The ledger keeps its newest
5,000 calls, so a long window on a busy install reaches back only as far as those rows do.

Example (/aux all, from the end-to-end test: one titling call, one compression call that
succeeded and three attempts of one that failed):

Auxiliary LLM calls, all recorded:

task              calls  failed  input  output  time
compression           4       3  1,000      50  0.0s
title_generation      1       0    321      12  1.1s
total                 5       3  1,321      62  1.1s

3 call(s) reported no token usage (streamed or failed).
  • calls counts provider attempts, so a retried call shows up once per attempt.
  • input is everything the provider read: fresh input plus cache reads and cache writes.
  • time is the wall time of the attempts.
  • Streamed and failed attempts carry no usage from Hermes; they are counted, with zero tokens,
    and the footer says how many there were.

/aux days puts one day on each line, newest day first, so the day a long session kept
compressing stands out. Days are your computer's local calendar days, and a day with no
auxiliary calls has no line. Example (the plugin's own output for a sample week):

Auxiliary LLM calls, last 7 days:

day         calls  failed    input  output   time
2026-10-06      5       1  106,105   5,250  1m17s
2026-10-05      1       0      290       9   0.7s
2026-10-04      2       0   41,310   2,111  19.2s
total           8       1  147,705   7,370  1m37s

1 call(s) reported no token usage (streamed or failed).

/aux recent lists the newest 10 calls of the window one by one (/aux recent 25 lists 25,
up to 100), newest first, so a slow or
failed compression a few minutes ago can be found without reading the totals. result is
ok or the exception class Hermes reported (never the error message). Example (the plugin's
own output for sample rows):

Auxiliary LLM calls, last 7 days, newest 5 of 5:

when                 task     model   input  output   time           result
1m ago   title_generation   m-small     321      12   1.1s               ok
39m ago       compression   m-small  40,310   1,990  21.2s               ok
40m ago       compression   m-small       0       0  30.0s  APITimeoutError
3h ago        compression   m-small  41,020   2,104  18.7s               ok
2d ago             vision  m-vision   1,580      64   3.4s               ok

What is recorded

One JSON line per attempt in <HERMES_HOME>/plugin-data/aux-ledger/calls.jsonl (per profile):

ts, session_id, aux_task, provider, model, response_model, retry_count,
streaming, duration_s, error_type, has_usage, input_tokens, output_tokens,
cache_read_tokens, cache_write_tokens, reasoning_tokens.

Not recorded: prompts, system prompts, responses, error messages (only the exception class
name), base URLs, API keys.

The file is bounded: once it passes 2 MB it is rewritten to its newest 5,000 rows.

Security and footprint

  • register() only registers one hook (post_auxiliary_call) and one command (/aux).
  • No network access, no subprocesses, no downloads, no credentials read.
  • Writes only calls.jsonl under its own plugin data directory. /aux clear deletes it.
  • Reads no Hermes state other than the hook payload it is handed.
  • The hook is observer-only and never raises: if the ledger cannot be written, the auxiliary
    call proceeds and the failure is logged at debug level.
  • Standard library only.

Development

python -m pytest tests                                   # unit tests, no Hermes needed
PYTHONPATH=<hermes checkout> python tests/e2e/hermes_e2e.py   # real loader + real auxiliary client
hermes plugins validate aux-ledger

Changes

  • 1.4.0: /aux recent takes a row count, /aux recent 25 (1 to 100; ten without
    one). What is recorded did not change.
  • 1.3.0: /aux recent, the newest 10 calls of any window, newest first, with their task,
    model, tokens, time and result. What is recorded did not change.
  • 1.2.0: /aux days, one line per day with auxiliary calls, newest day first; works with
    every window (/aux 30d days). The ledger also reads correctly after a Windows tool has saved
    it with a byte-order mark. What is recorded did not change.
  • 1.1.0: time windows (/aux 6h, /aux 7d, any number of hours or days) and grouping by
    provider (/aux providers). What is recorded did not change.
  • 1.0.0: first release.

MIT licensed.

README from the reviewed commit aee39a6; it updates when the author re-pins.

More by olric3015

18 plugins by olric3015 in the catalog.

olric3015approval-ledgerLocal ledger of Hermes's approval prompts (dangerous commands, protected writes, MCP consent, smart-mode verdicts): /approval-log shows how often each rule asked, how many prompts were approved, denied or left unanswered, and how long a person took to answer, per rule, per surface or per day, for the last 24 hours, any number of hours or days, the latest session or everything recorded; /approval-log recent lists the newest decisions (ten, or up to 100 with /approval-log recent 50). Disclosure — writes decision metadata (session id, surface, rule label, answer, timings; no command, description or session key) to plugin-data/approval-ledger/approvals.jsonl, trimmed to 5,000 rows past 2 MB; observer only, cannot answer or change an approval; no networkCommunitytoolsolric3015b64-ref@b64: context references that decode encoded text for the model: @b64:<value> takes base64 (standard or URL-safe, padded or not), hex (0x48656c6c6f, 48:65:6c, \x48\x65), percent-encoding or a data: URI, told apart by the value's shape or picked with base64:, base64url:, hex: or url:, and attaches what it holds: text as text with control and invisible characters escaped, JSON pretty-printed, gzip and zlib inflated in memory (at most 1 MiB, one layer), and binary named by its format (PNG and GIF with their size, JPEG, PDF, ZIP, ELF, SQLite and more) with a hex dump of its first 128 bytes; notes point out a value that was cut short, padding that is off, a second layer of encoding and a JWT inside; the TUI and Desktop composer preview a value as you type it. Disclosure — reads only the text of the reference itself, using Python's base64, json, urllib.parse and zlib; nothing decoded is run, opened or saved; no network, no files read or written, no subprocess, no credentials read; the plugin keeps no copy of what it decodes, but the @b64: reference and the decoded contents become part of the message sent to your model provider and saved in the session history, so decode only values you are willing to share with it; register() only adds the @b64: reference provider.Communitytoolsolric3015chmod-ref@chmod: context references that explain a file permission mode for the model: @chmod:<mode> takes an octal mode (755, 0755, 4755), ls -l text (-rwsr-xr-x, with the file type letter and a trailing + . or @) or an st_mode as stat and git write it (100644), and attaches both forms, each class's rights in words (directory rights named as such), what setuid, setgid and the sticky bit do and where they are ignored, and notes such as a mode everyone can change, one private to the owner, or a class holding rights the class before it lacks; @chmod:"644 u+x,go-w" applies a symbolic change step by step the way GNU chmod does (u g o a, + - =, r w x X s t, copies like u=g), @chmod:"100644 100755" shows what changes between two modes, and @chmod:"umask 027" shows what new files and directories get; the TUI and Desktop composer preview a mode as you type it. Disclosure — reads only the text of the reference itself, using Python's re; it never looks at a file on disk; no network, no files read or written, no permission changed, no subprocess, no credentials read; the plugin keeps no copy of what you write, but the @chmod: reference and the explanation become part of the message sent to your model provider and saved in the session history; register() only adds the @chmod: reference provider.Communitytoolsolric3015command-ledgerLocal ledger of the slash commands you run in Hermes (built-in commands in the CLI, built-in and plugin commands on messaging platforms): /command-log shows how often each command was used, its share, in how many sessions and when last, per command, platform, surface or day, for the last 24 hours, any number of hours or days, the latest session or everything recorded; /command-log recent lists the newest commands (ten, or up to 100 with /command-log recent 50) with the alias typed. Disclosure — writes command metadata (command name, alias typed, surface, platform, a 12-digit hash of the session key; no arguments, session key or user id) to plugin-data/command-ledger/commands.jsonl, trimmed to 5,000 rows past 2 MB; observer only, cannot block or change a command; no networkCommunitytools

Developer Mode

Explore built-in Hermes toolsets and actions.

HermesAgent

Developer?

Create a Hermes extension for Desktop or agent workflows.

Developer Section