chore(vault): capture generated Obsidian updates
This commit is contained in:
William Valentin
+16
@@ -0,0 +1,16 @@
|
||||
# Daily Hermes + AI Research Brief — 2026-05-21
|
||||
|
||||
## Important updates
|
||||
|
||||
- **No source-backed web brief could be produced today because this Hermes cron environment has no configured web search or extraction backend.** `web_search` returned “No web search provider configured,” and direct `web_extract` returned “No web extract provider configured. Set web.extract_backend to firecrawl, tavily, exa, or parallel.” Why it matters to Will: this daily research job cannot meet its current-news/source-backed requirement until the `web`/`search` tool backend is configured. Source targets that should be checked once fixed: [Hermes Agent GitHub releases](https://github.com/NousResearch/hermes-agent/releases), [Hermes Agent docs](https://hermes-agent.nousresearch.com/docs/), [Hermes Agent commits](https://github.com/NousResearch/hermes-agent/commits/main), [Model Context Protocol news](https://modelcontextprotocol.io/news), [GitHub changelog](https://github.blog/changelog/).
|
||||
|
||||
## Actionable ideas for us
|
||||
|
||||
- [quick] Configure a web backend for Hermes cron runs, e.g. set `web.extract_backend` to a supported provider and ensure the matching API key is available to the cron profile/session.
|
||||
- [quick] Run `hermes tools list` / `hermes tools` in the profile used by this scheduled job and confirm `web` or `search` is enabled for cron sessions.
|
||||
- [experiment] Add a preflight check to the brief job that reports missing web/search configuration distinctly before attempting research, so failures are obvious and not mistaken for “nothing notable.”
|
||||
- [watch] Tomorrow, search: `Hermes Agent NousResearch releases docs`, `AI agents MCP coding agents local LLM inference`, `model routing LLM ops privacy local-first AI agents`.
|
||||
|
||||
## Worth ignoring
|
||||
|
||||
- No hype/noise items were evaluated because live search/extraction was unavailable.
|
||||
+24
@@ -0,0 +1,24 @@
|
||||
# Daily Hermes + AI Research Brief — 2026-05-22
|
||||
|
||||
## Important updates
|
||||
|
||||
- **Hermes v0.14.0 is the practical upgrade to track.** The May 16 “Foundation Release” adds `pip install hermes-agent`, much lighter/lazier installs, ~19s faster cold start, 180× faster browser CDP calls, native Windows beta, LINE + SimpleX, Teams wiring, LSP semantic diagnostics on write, `/handoff`, and an OpenAI-compatible local proxy for OAuth providers. This directly helps Atlas/Hermes portability, cron startup latency, and interop with Codex/Aider/Cline/Continue. Source: [Hermes Agent releases](https://github.com/NousResearch/hermes-agent/releases).
|
||||
- **Hermes provider/routing surface keeps expanding.** v0.14 adds SuperGrok OAuth with grok-4.3 at 1M context, an OpenRouter Pareto Code router, first-class `x_search`, trusted `huggingface/skills` tap, and non-Anthropic `computer_use` via `cua-driver`. For Will: worth testing as alternate model routes for research/coding jobs without hard-coding one provider. Source: [Hermes Agent v0.14.0 release notes](https://github.com/NousResearch/hermes-agent/releases).
|
||||
- **Security checks are moving into MCP-native agent workflows.** GitHub made secret scanning in the GitHub MCP Server generally available, so MCP-compatible agents/IDEs can scan current changes for exposed secrets before commit/PR and honor existing push-protection customization. Source: [GitHub changelog](https://github.blog/changelog/2026-05-05-secret-scanning-with-github-mcp-server-is-now-generally-available/).
|
||||
- **Dependency scanning via MCP is now in public preview.** GitHub’s MCP Server `dependabot` toolset can return structured vulnerable-dependency results with package, severity, and fixed-version recommendations; it can also use Dependabot CLI locally for deeper post-commit diffing. This is a good pattern for Hermes tool/MCP guardrails around automated code edits. Source: [GitHub changelog](https://github.blog/changelog/2026-05-05-dependency-scanning-with-github-mcp-server-is-in-public-preview/).
|
||||
- **Codex is pushing long-running goal mode + shared plugin bundles.** OpenAI’s Codex changelog for 2026-05-21 says Goal mode is no longer experimental across app/IDE/CLI, Appshots can send Mac window context to Codex, remote computer use supports locked Macs, and Business plugin sharing can bundle skills, app integrations, and MCP servers. Useful competitive signal for Hermes `/goal`, skills, remote agents, and reusable bundles. Source: [Codex changelog](https://developers.openai.com/codex/changelog).
|
||||
- **vLLM’s Kubernetes docs remain a clean baseline for local/GPU agent infra.** The current deployment guide frames vLLM-on-K8s as a way to scale/manage ML models on GPU resources, with an OpenAI-compatible serving path elsewhere in the docs. This is relevant for Pi/local-swarm → GPU-cluster routing patterns. Source: [vLLM Kubernetes deployment docs](https://docs.vllm.ai/en/stable/deployment/k8s/).
|
||||
|
||||
## Actionable ideas for us
|
||||
|
||||
- [quick] Try `hermes proxy` locally as an OpenAI-compatible endpoint for one external coding tool, then note latency/auth rough edges.
|
||||
- [quick] Add a pre-commit/PR checklist item for “ask agent to run MCP secret/dependency scan” on Hermes/Atlas repos where GitHub MCP is configured.
|
||||
- [experiment] Compare Hermes `/goal` + Kanban against Codex Goal mode on the same long-running maintenance task; score recovery after restart, tool safety, and final diff quality.
|
||||
- [watch] Track Hermes v0.14.x patches for Windows beta, browser CDP speedups, `computer_use`, and Teams/LINE/SimpleX gateway regressions before relying on them in production workflows.
|
||||
|
||||
## Worth ignoring
|
||||
|
||||
- Generic “agent ecosystem maps” and listicles unless they include concrete protocol/tooling changes or reproducible benchmarks.
|
||||
- Consumer-only app demos around screenshots/mobile control unless they translate into reusable agent context-capture patterns for Hermes.
|
||||
|
||||
Saved to Obsidian: Atlas/Daily Research/2026-05-22 - Hermes AI Brief.md
|
||||
+165
@@ -0,0 +1,165 @@
|
||||
---
|
||||
title: Hermes Setup Video Follow-up Audit
|
||||
date: 2026-05-21
|
||||
tags:
|
||||
- atlas
|
||||
- hermes
|
||||
- operations
|
||||
- memory-hygiene
|
||||
- profiles
|
||||
- skills
|
||||
---
|
||||
|
||||
# Hermes Setup Video Follow-up Audit
|
||||
|
||||
Source: YouTube video `NvakBZyc1Sg`, summarized in Discord thread by Wrack/Atlas.
|
||||
|
||||
Wrack requested follow-ups: **1, 2, 4, 5** from the video-derived action list:
|
||||
|
||||
1. Audit compression threshold / auxiliary model config.
|
||||
2. Compact static persistent memory because it was near capacity.
|
||||
4. Ensure profile-local credentials remain synced for specialist workers.
|
||||
5. Keep turning recurring workflows into skills rather than memory bloat.
|
||||
|
||||
## Result
|
||||
|
||||
Completed on 2026-05-21.
|
||||
|
||||
## 1. Compression and auxiliary/fallback config
|
||||
|
||||
Default profile was already aligned with the video's recommendation to compress before context reaches 100%.
|
||||
|
||||
Observed default profile config:
|
||||
|
||||
```yaml
|
||||
model:
|
||||
default: gpt-5.5
|
||||
provider: openai-codex
|
||||
|
||||
fallback_providers:
|
||||
- provider: zai
|
||||
model: glm-5.1
|
||||
base_url: https://api.z.ai/api/paas/v4
|
||||
|
||||
compression:
|
||||
enabled: true
|
||||
threshold: 0.5
|
||||
target_ratio: 0.2
|
||||
rollover:
|
||||
enabled: true
|
||||
auto_rollover: true
|
||||
threshold: 0.5
|
||||
|
||||
session_rollover:
|
||||
enabled: true
|
||||
threshold: 0.5
|
||||
|
||||
auxiliary:
|
||||
compression:
|
||||
provider: zai
|
||||
model: glm-5.1
|
||||
```
|
||||
|
||||
Specialist profile summary:
|
||||
|
||||
| Profile | Primary provider/model | Compression threshold | Aux compression |
|
||||
|---|---|---:|---|
|
||||
| default | openai-codex / gpt-5.5 | 0.5 | zai / glm-5.1 |
|
||||
| engineer | openai-codex / gpt-5.5 | 0.5 | zai / glm-5.1 |
|
||||
| reviewer | openai-codex / gpt-5.5 | 0.5 | zai / glm-5.1 |
|
||||
| researcher | openai-codex / gpt-5.5 | 0.5 | zai / glm-5.1 |
|
||||
| writer | openai-codex / gpt-5.5 | 0.5 | zai / glm-5.1 |
|
||||
| ops | zai / glm-5.1 | 0.5 | zai / glm-5.1 |
|
||||
| orchestrator | zai / glm-5.1 | 0.5 | zai / glm-5.1 |
|
||||
| glm-simple | zai / glm-5.1 | 0.5 | zai / glm-5.1 |
|
||||
|
||||
No config changes were needed for compression.
|
||||
|
||||
## 2. Static memory compaction
|
||||
|
||||
Persistent memory and user profile were near capacity. A small compaction pass was completed without intentionally dropping durable facts.
|
||||
|
||||
Compacted entries included:
|
||||
|
||||
- Atlas/Hermes repo authority/remotes policy.
|
||||
- Obsidian vault and Hermes/AI brief location.
|
||||
- Obsidian autostart / Local REST API note.
|
||||
- CoreWeave/FMLA/benefits entry.
|
||||
- Family entry.
|
||||
|
||||
Approximate post-compaction state reported by the memory tool:
|
||||
|
||||
- General memory: about **95%** full.
|
||||
- User profile: about **96%** full.
|
||||
|
||||
Guideline reaffirmed: static memory should store durable facts only; procedures belong in skills, project notes, or session history.
|
||||
|
||||
## 4. Profile-local credential sync and smoke test
|
||||
|
||||
Before the audit, specialist profiles had no profile-local `auth.json` files.
|
||||
|
||||
The default profile's `~/.hermes/auth.json` was copied to:
|
||||
|
||||
- `~/.hermes/profiles/orchestrator/auth.json`
|
||||
- `~/.hermes/profiles/researcher/auth.json`
|
||||
- `~/.hermes/profiles/writer/auth.json`
|
||||
- `~/.hermes/profiles/ops/auth.json`
|
||||
- `~/.hermes/profiles/engineer/auth.json`
|
||||
- `~/.hermes/profiles/reviewer/auth.json`
|
||||
- `~/.hermes/profiles/glm-simple/auth.json`
|
||||
|
||||
Each file was installed with mode `600`.
|
||||
|
||||
Verified with `hermes --profile <profile> auth list` that each specialist profile could see the same provider credential set:
|
||||
|
||||
- `anthropic`
|
||||
- `copilot`
|
||||
- `openai-codex`
|
||||
- `zai`
|
||||
|
||||
Smoke-test flow:
|
||||
|
||||
```bash
|
||||
~/.hermes/scripts/specialist-profile-smoke-watchdog.sh
|
||||
```
|
||||
|
||||
First run reported a timeout for `ops`, but direct longer smoke test succeeded:
|
||||
|
||||
```bash
|
||||
timeout 180 hermes -p ops chat -q 'Reply exactly: ok' --toolsets safe -Q
|
||||
# -> ok
|
||||
```
|
||||
|
||||
Rerunning the watchdog was silent/successful.
|
||||
|
||||
## 5. Recurring workflows captured as a skill
|
||||
|
||||
Created local skill:
|
||||
|
||||
```text
|
||||
~/.hermes/skills/devops/recurring-hermes-workflows/SKILL.md
|
||||
```
|
||||
|
||||
Purpose: capture the durable workflow pattern from the video and Atlas practice:
|
||||
|
||||
- periodic report → Hermes cron;
|
||||
- health threshold alert → script-only cron, quiet on success;
|
||||
- durable multi-step project → Kanban;
|
||||
- repeatable procedure → skill plus scripts/references;
|
||||
- user feedback loop → compact durable preferences, not raw rating logs;
|
||||
- avoid adding long procedural recipes to always-injected memory.
|
||||
|
||||
This prevents future recurring-workflow logic from becoming static-memory bloat.
|
||||
|
||||
## Follow-up notes
|
||||
|
||||
- No urgent migration to VPS is recommended for the current Atlas setup.
|
||||
- Firecrawl/CamoFox remains optional; consider only when current search/browser stack hits Cloudflare or scraping limitations.
|
||||
- Re-run specialist profile smoke checks after auth/provider changes.
|
||||
- If memory pressure returns, compact or move procedural details into skills/Obsidian notes rather than increasing always-injected static memory.
|
||||
|
||||
## Related
|
||||
|
||||
- [[Atlas/Daily Research/2026-05-21 - Hermes AI Brief]]
|
||||
- Skill: `recurring-hermes-workflows`
|
||||
- Hermes skill reference: `hermes-agent/references/current-auth-fallback-and-profile-models.md`
|
||||
+34
@@ -0,0 +1,34 @@
|
||||
---
|
||||
type: daily-review
|
||||
date: 2026-05-21
|
||||
tags: [type/daily-review, automation/n8n]
|
||||
---
|
||||
|
||||
# Daily Review 2026-05-21
|
||||
|
||||
## Top priorities
|
||||
|
||||
- [ ]
|
||||
- [ ]
|
||||
- [ ]
|
||||
|
||||
## Inbox sweep
|
||||
|
||||
- [ ] Review [[Inbox]]
|
||||
- [ ] Promote useful captures into [[Projects Home]], [[Resources Home]], [[Decisions Home]], or [[Runbooks Home]]
|
||||
|
||||
## Open loops
|
||||
|
||||
- [ ] Check [[Projects Home]]
|
||||
- [ ] Check [[Meetings Home]] action items
|
||||
- [ ] Check [[Runbooks Home]] for procedures that need updates
|
||||
|
||||
## Notes / log
|
||||
|
||||
-
|
||||
|
||||
## End-of-day reflection
|
||||
|
||||
- What moved forward?
|
||||
- What is blocked?
|
||||
- What should start tomorrow?
|
||||
+34
@@ -0,0 +1,34 @@
|
||||
---
|
||||
type: daily-review
|
||||
date: 2026-05-22
|
||||
tags: [type/daily-review, automation/n8n]
|
||||
---
|
||||
|
||||
# Daily Review 2026-05-22
|
||||
|
||||
## Top priorities
|
||||
|
||||
- [ ]
|
||||
- [ ]
|
||||
- [ ]
|
||||
|
||||
## Inbox sweep
|
||||
|
||||
- [ ] Review [[Inbox]]
|
||||
- [ ] Promote useful captures into [[Projects Home]], [[Resources Home]], [[Decisions Home]], or [[Runbooks Home]]
|
||||
|
||||
## Open loops
|
||||
|
||||
- [ ] Check [[Projects Home]]
|
||||
- [ ] Check [[Meetings Home]] action items
|
||||
- [ ] Check [[Runbooks Home]] for procedures that need updates
|
||||
|
||||
## Notes / log
|
||||
|
||||
-
|
||||
|
||||
## End-of-day reflection
|
||||
|
||||
- What moved forward?
|
||||
- What is blocked?
|
||||
- What should start tomorrow?
|
||||
@@ -0,0 +1,30 @@
|
||||
# Inbox Triage 2026-05-21
|
||||
|
||||
Generated: 2026-05-21T01:00:00.054Z
|
||||
|
||||
## Inbox items
|
||||
|
||||
- [ ] [[Chat Summaries]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Inbox Home]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Inbox]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Weekly Review]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
|
||||
## Promote to Projects
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Resources
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Decisions
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Runbooks
|
||||
|
||||
- [ ]
|
||||
|
||||
## Archive / Delete / Defer
|
||||
|
||||
- [ ]
|
||||
@@ -0,0 +1,30 @@
|
||||
# Inbox Triage 2026-05-22
|
||||
|
||||
Generated: 2026-05-22T01:00:00.075Z
|
||||
|
||||
## Inbox items
|
||||
|
||||
- [ ] [[Chat Summaries]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Inbox Home]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Inbox]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Weekly Review]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
|
||||
## Promote to Projects
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Resources
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Decisions
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Runbooks
|
||||
|
||||
- [ ]
|
||||
|
||||
## Archive / Delete / Defer
|
||||
|
||||
- [ ]
|
||||
@@ -0,0 +1,30 @@
|
||||
# Inbox Triage 2026-05-23
|
||||
|
||||
Generated: 2026-05-23T01:00:00.042Z
|
||||
|
||||
## Inbox items
|
||||
|
||||
- [ ] [[Chat Summaries]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Inbox Home]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Inbox]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
- [ ] [[Weekly Review]] — classify as Project / Resource / Decision / Runbook / Archive
|
||||
|
||||
## Promote to Projects
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Resources
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Decisions
|
||||
|
||||
- [ ]
|
||||
|
||||
## Promote to Runbooks
|
||||
|
||||
- [ ]
|
||||
|
||||
## Archive / Delete / Defer
|
||||
|
||||
- [ ]
|
||||
+2
-2
@@ -1,12 +1,12 @@
|
||||
# Obsidian Automation Health
|
||||
|
||||
Updated: 2026-05-21T00:00:17.088Z
|
||||
Updated: 2026-05-23T01:00:49.446Z
|
||||
|
||||
## Status
|
||||
|
||||
- Overall: Needs attention
|
||||
- Obsidian REST: OK
|
||||
- Reindex trigger: {"error":{"message":"500 - \"{\\n \\\"error\\\": \\\"[Errno 2] No such file or directory: '/home/will/.hermes/skills/note-taking/rag-search/scripts/reindex_obsidian.sh'\\\"\\n}\"","name":"AxiosError","stack":"AxiosError: Request failed with status code 500\n at settle (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/axios@1.13.5/node_modules/axios/lib/core/settle.js:19:12)\n at RedirectableRequest.handleResponse (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/axios@1.13.5/node_modu
|
||||
- Reindex trigger: {"status":"ok","mode":"incremental","vault_path":"/home/will/lab/swarm/swarm-common/obsidian-vault/will/will-shared-zap","note_count":162,"added":0,"changed":0,"removed":0,"chunks_indexed":0,"vector_count":321,"state_file":"/home/will/.hermes/data/rag-search/obsidian_index_state.json","chroma_path":"/home/will/.hermes/data/rag-search/chroma"}
|
||||
- RAG/embedding check: {"ok":false,"status":"failed","exitCode":127,"output":"RAG health script missing: /home/will/.hermes/scripts/rag_embedding_health.py","durationMs":0}
|
||||
|
||||
This note is automatically overwritten by n8n.
|
||||
|
||||
Reference in New Issue
Block a user