lukemcqueen/hermes-cortex - Page 3

cross-agent-design

Before designing any cross-agent feature, protocol, or workflow: trace the receiving agent's end-to-end connectivity, auth, consumption, and permission path. This skill prevents the most expensive mistake in multi-agent systems — assuming that because Message…

Langue du texte source : anglais

fix-without-asking

When you discover an issue mid-task, the correct response is begin_change — not a question. Covers the litmus test, the failure pattern, and the concrete "never ask" guardrail. Triggers: 'should I fix', 'want me to', discovered anything wrong mid-task, user…

Langue du texte source : anglais

governance-sentinel

Recurring governance introspection — scrape brain sync snapshots for codepath patterns, compile weekly insights, and store verdicts in dream.json for retrospection. Runs as a sentinel after mycortex-update-sync, before daily police logs.

Langue du texte source : anglais

hermes-cortex-maintenance

Maintain an installed Hermes Cortex instance — update both the upstream Hermes Agent and the cortex repo layer, merge diverged histories, sync skills safely, and troubleshoot common issues. Covers the dual-update workflow (hermes update + cortex-update.sh),…

Langue du texte source : anglais

hermes-recovery

Server migration, disaster recovery, and restoration workflows for Hermes Agent — restoring from tar.gz archives, reconstructing from backup directories, and verifying all subsystems post-move.

Langue du texte source : anglais

loop-governance

TDD cycle scoring, self-improvement, and governance system for Hermes Cortex. Scores completeness/quality/progress per cycle with nomic embeddings, logs to SQLite, collects user feedback via CLI, auto-applies config patches with safety bounds, and integrates…

Langue du texte source : anglais

mycortex

Use for mycortex knowledge brain work or mycortex migration.

Langue du texte source : anglais

postgres-schema-design

Postgres schemas with RLS, roles, or migrations.

Langue du texte source : anglais

psql-automation

psql from scripts — error propagation, docker wrappers.

Langue du texte source : anglais

repo-health-review

Systematic repo health review — survey scripts, detect duplicates, check naming, find gaps, prune dead weight, and produce a structured report. Run before major refactors or when asked to audit the repo.

Langue du texte source : anglais

sensor-false-positive-remediation

Handle false positives from the auto-remediation sensor pipeline. Covers the trace-before-create workflow for sensor-reported missing scripts and distinguishing stale sensor entries from genuine issues.

Langue du texte source : anglais

server-hardening

Comprehensive security audit and hardening for Linux servers running web services (nginx, Docker, fail2ban, UFW). Five-layer methodology: network inventory → service config review → firewall/iptables audit → fail2ban/DDoS protection → OS sysctl hardening →…

Langue du texte source : anglais

orch-weekly-auto-fix

After the weekly opportunity scan identifies issues, run auto-fix patterns — git pull, branch cleanup, Docker restart, permission fixes, disk cleanup — then verify each fix succeeded. Reduces manual intervention for known recurring problems.

Langue du texte source : anglais

ollama-setup

Install, configure, and manage Ollama on Linux — including sudo-free tarball install, user systemd service, model management, and health verification.

Langue du texte source : anglais

save-lesson

Auto-save a bug-fix lesson after resolving any non-trivial error. Builds a personal bug-fix memory that makes the agent smarter over time. Trigger: after fixing a bug, after resolving an error, after debugging. Run: offline_knowledge lesson create --title…

Langue du texte source : anglais

agent-daily-bible-reading

Daily bible reading cron pattern — generates SOUL.md entries and brain pages for agent-wide scripture engagement.

Langue du texte source : anglais

cron-timezone-env

Cron timestamps follow HERMES_TIMEZONE env, never hardcoded.

Langue du texte source : anglais

file-ownership-boundaries

Know which files are yours to modify vs Hermes defaults. Covers the two-domain split (Hermes Agent vs hermes-cortex repo), what each domain owns, and how to check before editing.

Langue du texte source : anglais

agent-flow

Workflow router skill — classifies the incoming request into one of 12 patterns and dispatches to the correct tooling, output format, and checklist.

Langue du texte source : anglais

cron-job-management

Create, name, list, and maintain Hermes cron jobs — no_agent watchdog scripts, naming conventions, and the user's preferred listing format.

Langue du texte source : anglais

doc-size-budget

Slim docs to byte targets (SOUL 10K, AGENTS gates) safely.

Langue du texte source : anglais

cron-output-diagnostics

Use when a cron delivery is flagged or looks wrong.

Langue du texte source : anglais

staging-server-operations

Safe operational practices for Docker-based staging servers — volume management, change verification, and database recovery patterns.

Langue du texte source : anglais

cron-filesystem-fallback

Read cron job definitions and execution history from filesystem when cronjob CLI is unavailable. Platform-agnostic fallback for cron mode, restricted approval, or limited shell contexts.

Langue du texte source : anglais

bus-archive-diagnostics

Query bus queues and archives reliably for fleet results.

Langue du texte source : anglais

cron-cost-scheduling

Schedule LLM crons around provider peak/off-peak windows.

Langue du texte source : anglais

detached-worker-pattern

Cron tick budget kill: detached worker + result sweep.

Langue du texte source : anglais

mcp-health-monitoring

Probe MCP server health; extend the governance watchdog.

Langue du texte source : anglais

approval-gate-debugging

Use when a command is flagged as a security issue.

Langue du texte source : anglais

cortex-bus-automation

Automated Agent Bus processing via MCP.

Langue du texte source : anglais

cortex-bus-inbox

MCP inbox tools for cortex-bus messaging.

Langue du texte source : anglais

mcp-server-building

Build, test, and debug MCP servers for Hermes Agent — logging, dependency checks, fix hints, and best practices.

Langue du texte source : anglais

adversarial-finding-fix-patterns

Fix adversarial-verify.py findings with real handling.

Langue du texte source : anglais

yuanbao

Yuanbao (元宝) groups: @mention users, query info/members.

Langue du texte source : anglais

bus-queue-maintenance

Bus queue lifecycle — inspecting stuck messages, archiving orphaned messages, navigating state constraints, and performing routine queue cleanup. Complements cortex-bus for the hands-on maintenance side.

Langue du texte source : anglais

maintenance-scan

Systematic system health survey run proactively when the user gives an open-ended directive to "find work" or "look for issues". 9-step sequential scan: inbox → memory → doctor → crons → git → system → analyze → fix → verify.

Langue du texte source : anglais

documentation-scope

Multi-audience documentation scoping conventions for Hermes Cortex. Defines when and how to distinguish general (universal) content from deployment- specific content using scope banners and section tags. Keeps the public OSS repo usable by everyone while…

Langue du texte source : anglais

postgres-docker

Tune and configure PostgreSQL running inside Docker containers — custom configs, mounts, command overrides, and the listen_addresses pitfall.

Langue du texte source : anglais

repo-organization

Canonical repo organization for Hermes Cortex — structure, naming, consolidation, symlinks, and audit procedures.

Langue du texte source : anglais

nginx-web-app-deployment

Deploy a custom web app (Flask, Python, Node) behind nginx — upstream config, SSL, basic auth, rate limiting, launchd/systemd service, and multi-layer testing pattern.

Langue du texte source : anglais