wiki: sync 2026-10-22 03:03 UTC — 7 file(s) updated

This commit is contained in:
jarvis-at-skic committed 2026-10-22 03:03:07 +00:00
1 parent 444c0d7a34
commit 016681b6f8
7 files changed
+152 -32

No files matched your search

+87
View File
@@ -0,0 +1,87 @@
# Channel Wiki: #cuga-general
## Identity and purpose
- Guild: home-v1 (`1518726359512387766`).
- Category: skic-cuga (`1562373126288048188`).
- Channel: cuga-general (`1562373274032414780`).
- General CUGA coordination: discussion, questions, decisions, and next steps.
- Canonical upstream: https://github.com/cuga-project/cuga-agent (explicitly supplied by RootAtSkic). Private project copy: https://gitea.lego-cloud.eu/skic-cuga/cuga-agent.
## Setup authorization
RootAtSkic requested initial setup: https://discord.com/channels/1518726359512387766/1562373274032414780/1562373319074779178
## Workspace and continuity
- Workspace: `/opt/data/channels/cuga-general/`.
- Purpose folders: artifacts, research, repositories.
- Channel skill: `channel-general-skic-cuga`.
- Route source: `/opt/data/scripts/channel-wiki-sync.py`.
- Keep mutable context here rather than in config.yaml.
## Boundaries
Preserve existing privacy/permissions and mention behavior. No recurring automation, repository creation, deployment, or autonomous project execution is authorized by initial channel setup.
## Verified channel setup
- Topic set and read back.
- Guide pinned and verified: https://discord.com/channels/1518726359512387766/1562373274032414780/1562373981481476208
- Existing permission overwrites preserved.
## Remote skill repository
RootAtSkic authorized topology correction in message `1562379714742452256`. The existing repository (ID 325) was transferred to the private skills organization `skic-cuga-skills` and renamed, preserving history: https://gitea.lego-cloud.eu/skic-cuga-skills/channel-general-skic-cuga. The separate `skic-cuga` organization remains available for project repositories.
- Checkout: `/opt/data/channels/cuga-general/repositories/channel-general-skic-cuga`.
- Runtime: `/opt/data/skills/productivity/channel-general-skic-cuga`.
- Branch: `main`.
- Verified published commit: `98847b7c1e1902c563c82ecf011ae66aece80b73`.
- The skill explicitly documents both private Gitea organizations and their separate project-repository versus project-skill roles.
- Exact-commit authenticated remote SKILL.md equals checkout and runtime bytes; local and remote Git refs match.
- Authorization is specific to this organization and channel-skill repository; no application repositories or automation were created.
## CUGA import and comparison
RootAtSkic authorized importing CUGA into skic-cuga and producing extensive Hermes comparison as a Reveal.js HTML presentation, with Anime.js where useful (message `1562386610929016832`).
- Private independent import: https://gitea.lego-cloud.eu/skic-cuga/cuga-agent; not a scheduled mirror.
- Upstream main snapshot: `b219b48184c1544481e2568f2978ffb09fe8fbbb`; 246 branches and 45 tags verified equal to fetched upstream refs. Main default, README and LICENSE readback verified.
- Hermes upstream comparison snapshot: `8cf3a0b03d86a61c2ac9d50cd178acca86f6b564`; not an audit of the installed runtime.
- HTML: `/opt/data/channels/cuga-general/artifacts/CUGA-vs-Hermes.html`.
- Companion bundle: `/opt/data/channels/cuga-general/artifacts/CUGA-vs-Hermes-analysis.zip`.
- 29 narrative slides and 25 evidence slides; 50 qualified source claims. Detailed reports/ledgers under `research/`.
- Recommendation: retain Hermes as operator continuity, evaluate CUGA as a bounded complementary service. Static identity/policy-failure/execution-isolation concerns require controlled validation before exposure.
- Verification: immutable source locators and Git blobs checked; CUGA Python AST checks passed (syntax only). Browser navigation, offline loading, Anime.js replay, reduced-motion default and mobile reader exercised. No matched-model benchmark, product test suite, production deployment or exploit test performed.
## Revised technical-executive comparison
User message `1562398523239170128` requested 10–15 slides, challenged the implied absence of Hermes policy control, and authorized read-only login/inspection of https://jarvis-jr-v1.lego-cloud.eu.
- Corrected deliverable: `/opt/data/channels/cuga-general/artifacts/executive/CUGA-vs-Hermes-executive.html` — exactly 14 slides, no slide appendix; supersedes the earlier asymmetric policy framing.
- Evidence companion: `/opt/data/channels/cuga-general/artifacts/executive/CUGA-vs-Hermes-executive-evidence.zip`.
- Existing Keycloak account authenticated successfully. Inspected all 19 navigation routes and 11 targeted configuration sections. UI reports v0.20.2; exact installed commit not established. No configuration saves, job execution, task changes or credential reveals.
- Hermes supports general pre-tool veto/approval/argument rewriting, managed configuration, middleware and output transforms. Capabilities, configured state and tested enforcement are distinct.
- 13 CUGA advantage hypotheses challenged: 5 packaging differences, 3 overlaps, 3 unproven, 2 qualified advantages (versioned publication and integrated knowledge workflow). 41 exact code excerpts verified against immutable Git blobs.
- Console displayed approvals.mode=off and terminal.backend=local; unchanged, not runtime-tested. These are deployment settings, not missing product capabilities.
- HTML browser checks passed: 14 slides, no JavaScript errors/network loads/content overlap, navigation/animation/reduced-motion/mobile reading tested. Evidence ZIP rebuilt byte-identically. No product benchmark or production security certification.
## Audience-ready presentation
User message `1562404907091234829` requires the deck to stand alone for viewers with no conversation history.
- Current audience-facing deliverable: `/opt/data/channels/cuga-general/artifacts/standalone/CUGA-Hermes-Platform-Comparison.html` (14 slides).
- Narrative introduces AI agents and both products, sets selection criteria, compares management/governance/architecture/integration/knowledge/publication/security/cost, and closes with workload fit and a pilot recommendation.
- Removed correction/rebuttal language, internal project assumptions, private-console links and deployment-specific settings. Public evidence links and assessment limits remain.
- Revalidated all 14 slides: no JavaScript errors, external resource loads or content/footer overlap; navigation, animation and mobile report mode passed. Prior versions preserved.
## Outcome-coverage presentation
User message `1562407966366502945` requested a full reframing away from CUGA-favoring positioning and exactly three final appendix slides mapping CUGA features to investigated Hermes equivalents.
- Current deliverable: `/opt/data/channels/cuga-general/artifacts/outcomes/Hermes-CUGA-Outcome-Coverage.html` — 12 standalone narrative slides plus 3 final feature-mapping appendices, 15 slides total.
- New angle: can Hermes meet the required outcome, and what evidence would justify a second platform? Recommendation is to validate Hermes capabilities/integration paths before adoption decisions.
- Appendices map 15 selected features across guidance/controls, execution/operations, and knowledge/lifecycle. Each distinguishes built-in/setup, configuration, extension/integration and custom workflow. Versioned publication is explicitly a proposed external release workflow, not verified native publish-UI parity.
- Evidence companion: `/opt/data/channels/cuga-general/artifacts/outcomes/Feature-Mapping-Evidence.md`. Checked 26 immutable source links and 12 live documentation URLs; all source blobs/ranges matched and documentation returned 200.
- Browser verification passed for all 15 slides, offline loading, navigation, animation, reduced motion and mobile reading. No product runtime parity benchmark or deployment changes.
## Direct side-by-side comparison
User message `1562435905346609214` clarified that the goal is public-web-based evaluation/comparison, not migration or a Hermes-first adoption argument. Every main topic must show both products at comparable detail.
- Latest deliverable: `/opt/data/channels/cuga-general/artifacts/comparison/Hermes-vs-CUGA-Technical-Comparison.html`.
- 15 slides total: introduction plus 11 paired main-content slides, followed by the 3 previously requested feature-mapping appendices.
- Both developer surfaces are explicit: Hermes SDK-compatible APIs/plugins/tools/hooks versus CUGA Python SDK/providers. Architecture uses separately labeled product diagrams. Governance explains blocked calls, continued execution and approval-check errors in concrete terms.
- Paired topics: product overview, execution architecture, developer interfaces, rules/approvals/failure behavior, multi-agent coordination, memory/knowledge, management/publication, connectivity/automation, security, operating implications and comparison conclusions.
- Verified 39 distinct immutable source links against Git blobs/line ranges. Browser checks passed: all 15 slides fit; no JavaScript errors or external rendering requests; keyboard, contents, animation, mobile and reduced-motion behavior passed.
## Open context
No runtime adoption, production deployment or CUGA–Hermes integration has been authorized or implemented. The presentation proposes an isolated pilot with explicit acceptance gates.
## Source anchor — daily sync 2026-10-22
- Latest processed human message: `1562435905346609214`.
- First daily scan reconciled 16 human messages from full available channel history because no durable scan anchor existed. Current direction is a balanced public-web-based comparison, not migration or a Hermes-first adoption argument; preserve standalone audience context and three final mapping appendices. Requests and exact message IDs were verified live; import, presentation and browser-validation records above remain channel-workflow evidence, not independently re-executed by this sync.