Compare commits

...
Author SHA1 Message Date
william a2c00f6f8b Fix claude-bot replying to Hermes's own messages, route Hermes through LiteLLM
Critical bug: claude-bot only ignored its OWN messages and text explicitly
addressed to @hermes — it did not ignore Hermes's own replies appearing in
the room. A single @hermes mention cascaded into claude-bot replying to
Hermes's thread messages ('Hermes says: ...'), which could itself cascade
further. Fixed with an OTHER_AGENT_USER_IDS allowlist of sender IDs to
always ignore, not just a text-prefix check.

Also point Hermes's model provider at the local litellm gateway
(OPENAI_BASE_URL/OPENAI_API_KEY) instead of OpenRouter directly, matching
claude-agent's own chat path — one place to hold the OpenRouter credential.
This does NOT grant Hermes access to the Claude subscription; that's an
Anthropic-side restriction unrelated to which proxy sits in front of it,
already proven earlier in this session.

Separately (not a code fix): the 'Billing or credits exhausted: HTTP 402'
error Hermes hit is real — the OpenRouter account currently has 0 credits.
2026-08-23 16:02:15 +00:00
william c88fdcc2ea Merge pull request 'Fix Hermes: gateway command + disable network-reachable API server' (#11) from fix/hermes-gateway-command-and-api-server into main
Reviewed-on: #11
2026-08-23 15:52:35 +00:00
william 98762e764a Fix Hermes container: add gateway command, disable network-reachable API server
Without an explicit command the image launches the interactive CLI by
default, which immediately exits with 'Input is not a terminal' in a
detached container — it was doing nothing on every restart. Also disables
API_SERVER_ENABLED: Hermes itself warns at startup that a network-reachable
API server combined with the default unsandboxed 'local' terminal backend
gives any caller on the network full terminal/file access. Not needed yet
(Matrix is the actual interface) — can re-enable properly (with a sandboxed
terminal backend) if claude-agent ever needs to call Hermes programmatically.
2026-08-23 15:52:08 +00:00
william 5507192ee6 Merge pull request 'Add Hermes Agent as a second native Matrix presence' (#10) from feat/hermes-matrix into main
build-agent / build-and-push (push) Successful in 11s
Reviewed-on: #10
2026-08-23 15:49:05 +00:00
3 changed files with 37 additions and 9 deletions
+4
View File
@@ -30,6 +30,10 @@ MATRIX_CONTROL_ROOM_ID=
# Required: without it the bot can't tell its own messages apart from real ones and would # Required: without it the bot can't tell its own messages apart from real ones and would
# reply to itself in a loop, so it refuses to start. # reply to itself in a loop, so it refuses to start.
MATRIX_BOT_USER_ID= MATRIX_BOT_USER_ID=
# Comma-separated Matrix IDs of OTHER agents sharing the control room (currently just
# @hermes:...) — without this, claude-bot treats every message another bot posts as
# fresh chat input and replies to it, which that bot may then react to in turn.
OTHER_AGENT_USER_IDS=@hermes:matrix.apps.williamturner.eu
# Comma-separated "owner/repo" list the chat router is allowed to open code-change PRs # Comma-separated "owner/repo" list the chat router is allowed to open code-change PRs
# against. A plain chat message mentioning a repo NOT in this list is treated as chat, # against. A plain chat message mentioning a repo NOT in this list is treated as chat,
# never as a code task — the router only matches confidently against known repos. # never as a code task — the router only matches confidently against known repos.
+11
View File
@@ -11,6 +11,15 @@ const GITEA_URL = process.env.GITEA_URL;
// to the endpoint itself. This is also the only reliable way to filter the bot's own // to the endpoint itself. This is also the only reliable way to filter the bot's own
// messages now that there's no command prefix to naturally exclude them by. // messages now that there's no command prefix to naturally exclude them by.
const BOT_USER_ID = process.env.MATRIX_BOT_USER_ID; const BOT_USER_ID = process.env.MATRIX_BOT_USER_ID;
// Other agents sharing this room (currently just Hermes) — their own messages must be
// ignored the same way claude-bot ignores its own, or claude-bot's classifier treats
// every message another bot posts as fresh chat input and replies to it, which that
// bot may then react to in turn. Found the hard way: a single @hermes mention cascaded
// into claude-bot replying to Hermes's own thread messages ("Hermes says: ...").
const OTHER_AGENT_USER_IDS = (process.env.OTHER_AGENT_USER_IDS || "")
.split(",")
.map((id) => id.trim())
.filter(Boolean);
const KNOWN_REPOS = (process.env.KNOWN_REPOS || "") const KNOWN_REPOS = (process.env.KNOWN_REPOS || "")
.split(",") .split(",")
.map((r) => r.trim()) .map((r) => r.trim())
@@ -47,6 +56,7 @@ export async function startMatrixBot() {
client.on("room.message", async (roomId, event) => { client.on("room.message", async (roomId, event) => {
if (roomId !== CONTROL_ROOM_ID) return; if (roomId !== CONTROL_ROOM_ID) return;
if (event.sender === BOT_USER_ID) return; if (event.sender === BOT_USER_ID) return;
if (OTHER_AGENT_USER_IDS.includes(event.sender)) return;
const body = event.content?.body; const body = event.content?.body;
if (!body) return; if (!body) return;
// Messages explicitly addressed to another agent in this room (currently just // Messages explicitly addressed to another agent in this room (currently just
@@ -67,6 +77,7 @@ export async function startMatrixBot() {
} }
const reply = await chatReply(body); const reply = await chatReply(body);
console.log("chat reply sent, length:", reply.length);
await client.sendText(roomId, truncate(reply)); await client.sendText(roomId, truncate(reply));
} catch (err) { } catch (err) {
console.error("message handling failed", err); console.error("message handling failed", err);
+22 -9
View File
@@ -77,16 +77,29 @@ services:
# rooms (DMs to it would respond unprompted, per Hermes's own default behavior). # rooms (DMs to it would respond unprompted, per Hermes's own default behavior).
MATRIX_ALLOWED_USERS: ${MATRIX_HUMAN_USER_ID} MATRIX_ALLOWED_USERS: ${MATRIX_HUMAN_USER_ID}
MATRIX_REQUIRE_MENTION: "true" MATRIX_REQUIRE_MENTION: "true"
OPENROUTER_API_KEY: ${OPENROUTER_API_KEY} # Routed through the local litellm gateway, not OpenRouter directly — same pattern
# Exposed on the internal network only (see claude-agent's LITELLM_BASE_URL-style # as claude-agent's chat path, one place to hold the OpenRouter credential and swap
# usage pattern) — nothing publishes this port externally. # models. Hermes's "main"/custom-endpoint provider is any OpenAI-compatible API
API_SERVER_ENABLED: "true" # reachable via OPENAI_BASE_URL + OPENAI_API_KEY. Note: this does NOT give Hermes
API_SERVER_HOST: 0.0.0.0 # access to the Claude Pro/Max subscription — that's blocked by Anthropic itself for
API_SERVER_KEY: ${HERMES_API_SERVER_KEY} # any caller other than the real Claude Code CLI, proven earlier in this session
# (reproduced with plain curl straight to api.anthropic.com, LiteLLM or not).
OPENAI_BASE_URL: http://litellm:4000/v1
OPENAI_API_KEY: ${LITELLM_MASTER_KEY}
# Left disabled: Hermes itself warns that a network-reachable API server combined
# with the default unsandboxed ('local') terminal backend gives any caller full
# terminal/file access within the container. Matrix is the actual interface in use;
# re-enable (API_SERVER_HOST: 0.0.0.0) only alongside terminal.backend: docker if
# claude-agent ever needs to call Hermes programmatically.
API_SERVER_ENABLED: "false"
volumes: volumes:
- /home/william/hermes-data:/opt/data - /home/william/hermes-data:/opt/data
networks: networks:
- web - web
# Without this the image's default command launches the interactive CLI, which
# immediately exits ("Input is not a terminal") since a detached container has no
# stdin — the container then just sits there having done nothing, every restart.
command: ["gateway", "run"]
claude-agent: claude-agent:
image: ${GITEA_REGISTRY_IMAGE} image: ${GITEA_REGISTRY_IMAGE}
@@ -108,10 +121,10 @@ services:
MATRIX_BOT_TOKEN: ${MATRIX_BOT_TOKEN} MATRIX_BOT_TOKEN: ${MATRIX_BOT_TOKEN}
MATRIX_CONTROL_ROOM_ID: ${MATRIX_CONTROL_ROOM_ID} MATRIX_CONTROL_ROOM_ID: ${MATRIX_CONTROL_ROOM_ID}
MATRIX_BOT_USER_ID: ${MATRIX_BOT_USER_ID} MATRIX_BOT_USER_ID: ${MATRIX_BOT_USER_ID}
OTHER_AGENT_USER_IDS: ${OTHER_AGENT_USER_IDS}
KNOWN_REPOS: ${KNOWN_REPOS} KNOWN_REPOS: ${KNOWN_REPOS}
# All model calls now go through the local litellm service, not OpenRouter directly — # Chat replies go through the local litellm service (OpenRouter's models, incl. its
# one gateway for OpenRouter's models (incl. its auto-router) and, for the # auto-router), not OpenRouter directly.
# claude-subscription route, Anthropic itself via the forwarded OAuth token above.
LITELLM_BASE_URL: http://litellm:4000 LITELLM_BASE_URL: http://litellm:4000
LITELLM_MASTER_KEY: ${LITELLM_MASTER_KEY} LITELLM_MASTER_KEY: ${LITELLM_MASTER_KEY}
volumes: volumes: